You might have a nvidia video card on your motherboard. There are two choices here. Try to use the nvidia or turn it off and plug in your old known video card. Today I wish I had done the latter because using nvidia with f7 is a pain.
I really do not see a new Linux user ever getting his/her computer working with nvidia. You need to go to the nvidia web page and get a tarball and install it, not a new person's thing, or you can get 4 rpm files and learn to use --nodeps at the proper time.
Hoping that the updates would by now have some nvidia help, after getting 236 updates last night on my f7-64 bit system it did not fix the problem. I used the 4 rpm files from www.atrpms.net which worked but maybe not well. I heard from Ric Moore that the tarball is the way to go. I will try that on f8.
A bug I keep forgetting to file is the following. A really bad problem with nvidia is the missing pointer when X windows boots up. You can do nothing! This is fixed by edit of the /etc/X11/xrog.conf file adding you want to use a software pointer.
But this will not work if grub.conf has a kernel directive to use rhxxx which hides the boot up output. While that standard kernel directive exists you can not get a pointer period.
This bug makes f7 and I expect f8 useless to a new user with nvidia.
On Sun, 2007-10-28 at 06:44 -0600, Karl Larsen wrote:
You might have a nvidia video card on your motherboard. There are two choices here. Try to use the nvidia or turn it off and plug in your old known video card. Today I wish I had done the latter because using nvidia with f7 is a pain.
That depends on which card you have. I've two here that were easy enough. One fairly old, one quite new.
I really do not see a new Linux user ever getting his/her computer
working with nvidia. You need to go to the nvidia web page and get a tarball and install it, not a new person's thing, or you can get 4 rpm files and learn to use --nodeps at the proper time.
No, you don't. Well, maybe *you* do, but not everybody. I enabled the Livna repo (which I use for more than just nvidia), yum installed kmod-nvidia-something_or_other, and that was virtually it.
Some older cards may be a bit more of a hassle. Some other cards from other companies may be a total impossibility.
A bug I keep forgetting to file is the following. A really bad
problem with nvidia is the missing pointer when X windows boots up. You can do nothing! This is fixed by edit of the /etc/X11/xrog.conf file adding you want to use a software pointer.
But this will not work if grub.conf has a kernel directive to use
rhxxx which hides the boot up output. While that standard kernel directive exists you can not get a pointer period.
This bug makes f7 and I expect f8 useless to a new user with nvidia.
RHGB is not compulsory, and the pointer bug only exists with some cards. It doesn't with mine.
Tim wrote:
On Sun, 2007-10-28 at 06:44 -0600, Karl Larsen wrote:
You might have a nvidia video card on your motherboard. There are two choices here. Try to use the nvidia or turn it off and plug in your old known video card. Today I wish I had done the latter because using nvidia with f7 is a pain.
That depends on which card you have. I've two here that were easy enough. One fairly old, one quite new.
I have one that is VERY HARD. You see if you can get the name of your cards. I will look for mine. We can then steer Linux users away from mine.
I really do not see a new Linux user ever getting his/her computer
working with nvidia. You need to go to the nvidia web page and get a tarball and install it, not a new person's thing, or you can get 4 rpm files and learn to use --nodeps at the proper time.
No, you don't. Well, maybe *you* do, but not everybody. I enabled the Livna repo (which I use for more than just nvidia), yum installed kmod-nvidia-something_or_other, and that was virtually it.
I do too have Livna eneabled so I can get VLC the best image and sound device on Linux. I also have 2 kmod-nvidia... files on my computer now.
Some older cards may be a bit more of a hassle. Some other cards from other companies may be a total impossibility.
We are talking here about a new card on a new motherboard that is a nvidia card. It stinks!
A bug I keep forgetting to file is the following. A really bad
problem with nvidia is the missing pointer when X windows boots up. You can do nothing! This is fixed by edit of the /etc/X11/xrog.conf file adding you want to use a software pointer.
But this will not work if grub.conf has a kernel directive to use
rhxxx which hides the boot up output. While that standard kernel directive exists you can not get a pointer period.
This bug makes f7 and I expect f8 useless to a new user with nvidia.
RHGB is not compulsory, and the pointer bug only exists with some cards. It doesn't with mine.
I see, all 10 of your nvidia equipt computers have never shown you a missing pointer? Amazing.
On 28/10/2007, Karl Larsen k5di@zianet.com wrote:
You might have a nvidia video card on your motherboard. There are
two choices here. Try to use the nvidia or turn it off and plug in your old known video card. Today I wish I had done the latter because using nvidia with f7 is a pain.
I really do not see a new Linux user ever getting his/her computer
working with nvidia. You need to go to the nvidia web page and get a tarball and install it, not a new person's thing, or you can get 4 rpm files and learn to use --nodeps at the proper time.
Firstly, Fedora will work out of the box with nvidia cards using the free/OSS drivers. They may not yet properly support 3D, but they do work and give you a graphical interface.
At that point, if you do want the extra 3D glits, installing the proprietory NVidia drivers is as trivial as this:
As root: 1) rpm -ivh http://rpm.livna.org/livna-release-7.rpm 2) rpm --import /etc/pki/rpm-gpg/RPM-GPG-KEY-livna 3) yum install kmod-nvidia
That is ALL there is to it. You're making your life overly difficult.
Tim:
That depends on which card you have. I've two here that were easy enough. One fairly old, one quite new.
I have one that is VERY HARD. You see if you can get the name of
your cards. I will look for mine. We can then steer Linux users away from mine.
I'll have to check on the other ones later, but this one that went off without a hitch, is a: Quadro NVS 110M / GeForce Go 7300 rev 161
That was on a new laptop, too. Which can be pushing one's luck.
Karl:
This bug makes f7 and I expect f8 useless to a new user with nvidia.
RHGB is not compulsory, and the pointer bug only exists with some cards. It doesn't with mine.
I see, all 10 of your nvidia equipt computers have never shown you a
missing pointer? Amazing.
Try turning that around. YOUR experiences with YOUR cards are just YOUR experiences. Just because YOU'VE had some trouble with YOUR cards doesn't mean that FC 7 or 8 are going to be "useless" with NVidia cards for other people.
If I've told you once, I've told you a thousand times, stop exaggerating. :-\
Jonathan Underwood wrote:
On 28/10/2007, Karl Larsen k5di@zianet.com wrote:
You might have a nvidia video card on your motherboard. There are
two choices here. Try to use the nvidia or turn it off and plug in your old known video card. Today I wish I had done the latter because using nvidia with f7 is a pain.
I really do not see a new Linux user ever getting his/her computer
working with nvidia. You need to go to the nvidia web page and get a tarball and install it, not a new person's thing, or you can get 4 rpm files and learn to use --nodeps at the proper time.
Firstly, Fedora will work out of the box with nvidia cards using the free/OSS drivers. They may not yet properly support 3D, but they do work and give you a graphical interface.
At that point, if you do want the extra 3D glits, installing the proprietory NVidia drivers is as trivial as this:
As root:
- rpm -ivh http://rpm.livna.org/livna-release-7.rpm
- rpm --import /etc/pki/rpm-gpg/RPM-GPG-KEY-livna
- yum install kmod-nvidia
That is ALL there is to it. You're making your life overly difficult.
With those supplied with F7 my computer would not show a full screen but one offset by about 70 degrees. And out of the box it has no pointer on X windows. What is my computer?
Well it is a SY-P4VGM v1.0 motherboard which has a nvidia video card undefined in the small book they provide. A CD-Rom is included and I will look at that. That is a SOYO motherboard at www,soyousa.com and I will look for the nvidio name there.
Tim wrote:
Tim:
That depends on which card you have. I've two here that were easy enough. One fairly old, one quite new.
I have one that is VERY HARD. You see if you can get the name of
your cards. I will look for mine. We can then steer Linux users away from mine.
I'll have to check on the other ones later, but this one that went off without a hitch, is a: Quadro NVS 110M / GeForce Go 7300 rev 161
That was on a new laptop, too. Which can be pushing one's luck.
Karl:
This bug makes f7 and I expect f8 useless to a new user with nvidia.
RHGB is not compulsory, and the pointer bug only exists with some cards. It doesn't with mine.
I see, all 10 of your nvidia equipt computers have never shown you a
missing pointer? Amazing.
Try turning that around. YOUR experiences with YOUR cards are just YOUR experiences. Just because YOU'VE had some trouble with YOUR cards doesn't mean that FC 7 or 8 are going to be "useless" with NVidia cards for other people.
If I've told you once, I've told you a thousand times, stop exaggerating. :-\
Is the livna depository going to be standard for all who get F8? If not who will tell them they need to set it up and how to do that?
On Sunday 28 October 2007 11:35:34 am Karl Larsen wrote:
Jonathan Underwood wrote:
On 28/10/2007, Karl Larsen k5di@zianet.com wrote:
You might have a nvidia video card on your motherboard. There are
two choices here. Try to use the nvidia or turn it off and plug in your old known video card. Today I wish I had done the latter because using nvidia with f7 is a pain.
I really do not see a new Linux user ever getting his/her computer
working with nvidia. You need to go to the nvidia web page and get a tarball and install it, not a new person's thing, or you can get 4 rpm files and learn to use --nodeps at the proper time.
Firstly, Fedora will work out of the box with nvidia cards using the free/OSS drivers. They may not yet properly support 3D, but they do work and give you a graphical interface.
At that point, if you do want the extra 3D glits, installing the proprietory NVidia drivers is as trivial as this:
As root:
- rpm -ivh http://rpm.livna.org/livna-release-7.rpm
- rpm --import /etc/pki/rpm-gpg/RPM-GPG-KEY-livna
- yum install kmod-nvidia
That is ALL there is to it. You're making your life overly difficult.
With those supplied with F7 my computer would not show a full screen
but one offset by about 70 degrees. And out of the box it has no pointer on X windows. What is my computer?
Well it is a SY-P4VGM v1.0 motherboard which has a nvidia video
card undefined in the small book they provide. A CD-Rom is included and I will look at that. That is a SOYO motherboard at www,soyousa.com and I will look for the nvidio name there.
Karl: According to the Soyo site, your motherboard has an onboard Prosavage graphics chip (http://www.soyousa.com/products/proddesc.php?t=d&id=292), and uses a VIA chipset. Nothing there about nVidia, so I assume that the nVidia card is an add-on. Is it possible that there is a conflict between the onboard chip and the nVidia card? In particular, has the onboard device been disabled in the BIOS?
-- cmg (who is very wary on on-board stuff)
On Sun, 2007-10-28 at 12:00 -0400, Carroll Grigsby wrote:
According to the Soyo site, your motherboard has an onboard Prosavage graphics chip (http://www.soyousa.com/products/proddesc.php?t=d&id=292), and uses a VIA chipset. Nothing there about nVidia, so I assume that the nVidia card is an add-on.
/sbin/lspci probably will tell.
Ralf
On Sun, 2007-10-28 at 14:21 +0000, Jonathan Underwood wrote:
On 28/10/2007, Karl Larsen k5di@zianet.com wrote:
You might have a nvidia video card on your motherboard. There are
two choices here. Try to use the nvidia or turn it off and plug in your old known video card. Today I wish I had done the latter because using nvidia with f7 is a pain.
I really do not see a new Linux user ever getting his/her computer
working with nvidia. You need to go to the nvidia web page and get a tarball and install it, not a new person's thing, or you can get 4 rpm files and learn to use --nodeps at the proper time.
Firstly, Fedora will work out of the box with nvidia cards using the free/OSS drivers. They may not yet properly support 3D, but they do work and give you a graphical interface.
Well, mostly...
https://bugzilla.redhat.com/show_bug.cgi?id=249367
At that point, if you do want the extra 3D glits, installing the proprietory NVidia drivers is as trivial as this:
As root:
- rpm -ivh http://rpm.livna.org/livna-release-7.rpm
- rpm --import /etc/pki/rpm-gpg/RPM-GPG-KEY-livna
- yum install kmod-nvidia
That is ALL there is to it. You're making your life overly difficult.
The proprietary drivers don't work with my laptop card either. The brightness controls apparently don't communicate with the driver correctly and suspend/hibernate/resume results in black screens or lockups.
Even the vesa driver doesn't work perfectly. The backlight doesn't go off and suspend doesn't work.
The machine won't suspend at all, apperntly due to some interaction with ACPI and the video card.
In the past, I have found that Nvidia has been pretty good compared with other manufacturers (especially ATI) about their proprietary drivers. Now that ATI is opening their drivers, I will be seriously considering them in the future. I'm stuck with the laptop for a while, though 8^(.
Carroll Grigsby wrote:
On Sunday 28 October 2007 11:35:34 am Karl Larsen wrote:
Jonathan Underwood wrote:
On 28/10/2007, Karl Larsen k5di@zianet.com wrote:
You might have a nvidia video card on your motherboard. There are
two choices here. Try to use the nvidia or turn it off and plug in your old known video card. Today I wish I had done the latter because using nvidia with f7 is a pain.
I really do not see a new Linux user ever getting his/her computer
working with nvidia. You need to go to the nvidia web page and get a tarball and install it, not a new person's thing, or you can get 4 rpm files and learn to use --nodeps at the proper time.
Firstly, Fedora will work out of the box with nvidia cards using the free/OSS drivers. They may not yet properly support 3D, but they do work and give you a graphical interface.
At that point, if you do want the extra 3D glits, installing the proprietory NVidia drivers is as trivial as this:
As root:
- rpm -ivh http://rpm.livna.org/livna-release-7.rpm
- rpm --import /etc/pki/rpm-gpg/RPM-GPG-KEY-livna
- yum install kmod-nvidia
That is ALL there is to it. You're making your life overly difficult.
With those supplied with F7 my computer would not show a full screen
but one offset by about 70 degrees. And out of the box it has no pointer on X windows. What is my computer?
Well it is a SY-P4VGM v1.0 motherboard which has a nvidia video
card undefined in the small book they provide. A CD-Rom is included and I will look at that. That is a SOYO motherboard at www,soyousa.com and I will look for the nvidio name there.
Karl: According to the Soyo site, your motherboard has an onboard Prosavage graphics chip (http://www.soyousa.com/products/proddesc.php?t=d&id=292), and uses a VIA chipset. Nothing there about nVidia, so I assume that the nVidia card is an add-on. Is it possible that there is a conflict between the onboard chip and the nVidia card? In particular, has the onboard device been disabled in the BIOS?
-- cmg (who is very wary on on-board stuff)
Good information. My book says nothing about VGA except to say it's there. On the companion CD-Rom I found that the drivers are via-km400 which do sound like prosavage which is well known to Linux.
But for the fact that I get good video ONLY when nvidia software is correct and in place. So there must be some hardware somewhere also on the motherboard.
On Sun October 28 2007, Karl Larsen wrote:
Is the livna depository going to be standard for all who get F8? If not who will tell them they need to set it up and how to do that?
Livna will never be a 'standard' include with Fedora, Karl. It contains packages which some lawyers argue are in violation of U.S. Copyright laws. If Fedora were to include links to such a repository in their distribution, they could open themselves up to lawsuits, since Fedora is U.S. based.
Ralf Corsepius wrote:
On Sun, 2007-10-28 at 12:00 -0400, Carroll Grigsby wrote:
According to the Soyo site, your motherboard has an onboard Prosavage graphics chip (http://www.soyousa.com/products/proddesc.php?t=d&id=292), and uses a VIA chipset. Nothing there about nVidia, so I assume that the nVidia card is an add-on.
/sbin/lspci probably will tell.
Ralf
Thanks Ralf, you are right again. Here is a whole list of nvidia hardware:
[root@k5di ~]# lspci 00:00.0 RAM memory: nVidia Corporation C51 Host Bridge (rev a2) 00:00.1 RAM memory: nVidia Corporation C51 Memory Controller 0 (rev a2) 00:00.2 RAM memory: nVidia Corporation C51 Memory Controller 1 (rev a2) 00:00.3 RAM memory: nVidia Corporation C51 Memory Controller 5 (rev a2) 00:00.4 RAM memory: nVidia Corporation C51 Memory Controller 4 (rev a2) 00:00.5 RAM memory: nVidia Corporation C51 Host Bridge (rev a2) 00:00.6 RAM memory: nVidia Corporation C51 Memory Controller 3 (rev a2) 00:00.7 RAM memory: nVidia Corporation C51 Memory Controller 2 (rev a2) 00:03.0 PCI bridge: nVidia Corporation C51 PCI Express Bridge (rev a1) 00:04.0 PCI bridge: nVidia Corporation C51 PCI Express Bridge (rev a1) 00:05.0 VGA compatible controller: nVidia Corporation C51G [GeForce 6100] (rev a2) 00:09.0 RAM memory: nVidia Corporation MCP51 Host Bridge (rev a2) 00:0a.0 ISA bridge: nVidia Corporation MCP51 LPC Bridge (rev a3) 00:0a.1 SMBus: nVidia Corporation MCP51 SMBus (rev a3) 00:0a.2 RAM memory: nVidia Corporation MCP51 Memory Controller 0 (rev a3) 00:0b.0 USB Controller: nVidia Corporation MCP51 USB Controller (rev a3) 00:0b.1 USB Controller: nVidia Corporation MCP51 USB Controller (rev a3) 00:0d.0 IDE interface: nVidia Corporation MCP51 IDE (rev a1) 00:0e.0 IDE interface: nVidia Corporation MCP51 Serial ATA Controller (rev a1) 00:10.0 PCI bridge: nVidia Corporation MCP51 PCI Bridge (rev a2) 00:10.2 Multimedia audio controller: nVidia Corporation MCP51 AC97 Audio Controller (rev a2) 00:14.0 Bridge: nVidia Corporation MCP51 Ethernet Controller (rev a3) 00:18.0 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] HyperTransport Technology Configuration 00:18.1 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Address Map 00:18.2 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] DRAM Controller 00:18.3 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Miscellaneous Control [root@k5di ~]#
And the video controller is a nVidia Corporation C51G [GeForce 6100] (rev a2) what ever that might mean. Also the entire computer is full of nvidia things and the only one causing trouble is the video.
On 28/10/2007, Matthew Saltzman mjs@clemson.edu wrote:
In the past, I have found that Nvidia has been pretty good compared with other manufacturers (especially ATI) about their proprietary drivers. Now that ATI is opening their drivers, I will be seriously considering them in the future. I'm stuck with the laptop for a while, though 8^(.
At this point in time I have purged my life of all nvidia cards in favour of intel which have OSS drivers contributed by Intel. They may not be high end gaming cards, but they work damn well under linux, and so I'd recommend them highly if that is an option for you.
Claude Jones wrote:
On Sun October 28 2007, Karl Larsen wrote:
Is the livna depository going to be standard for all who get F8? If
not who will tell them they need to set it up and how to do that?
Livna will never be a 'standard' include with Fedora, Karl. It contains packages which some lawyers argue are in violation of U.S. Copyright laws. If Fedora were to include links to such a repository in their distribution, they could open themselves up to lawsuits, since Fedora is U.S. based.
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
On Sun October 28 2007, Les Mikesell wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
I think any but the most recent members of this list have noted your position on that subject, Les --- it would have been hard to miss '>
Claude Jones wrote:
On Sun October 28 2007, Les Mikesell wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
I think any but the most recent members of this list have noted your position on that subject, Les --- it would have been hard to miss '>
And yet, people still try to run fedora on machines with nvidia and seem surprised by the problems...
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
On Sun, Oct 28, 2007 at 12:24:52PM -0400, Claude Jones wrote:
On Sun October 28 2007, Karl Larsen wrote:
Is the livna depository going to be standard for all who get F8? If not who will tell them they need to set it up and how to do that?
Livna will never be a 'standard' include with Fedora, Karl. It contains packages which some lawyers argue are in violation of U.S. Copyright laws.
I think its actually US patent laws that are the problem.
On Sun, 2007-28-10 at 18:51 +0000, Dave Mitchell wrote:
On Sun, Oct 28, 2007 at 12:24:52PM -0400, Claude Jones wrote:
On Sun October 28 2007, Karl Larsen wrote:
Is the livna depository going to be standard for all who get F8? If
not who will tell them they need to set it up and how to do that?
Livna will never be a 'standard' include with Fedora, Karl. It contains packages which some lawyers argue are in violation of U.S. Copyright laws.
I think its actually US patent laws that are the problem.
--
According to Max Spevack, the Fedora chair, the project is legally part of Red Hat, and must follow US import laws. To protect itself, and at the advice of Red Hat's lawyers, the project cannot even refer in release notes to repositories like livna that contain packages of dubious legality such as lame. If it did, it might technically run afoul of those import laws.
Firstly, Fedora will work out of the box with nvidia cards using the free/OSS drivers. They may not yet properly support 3D, but they do work and give you a graphical interface.
At that point, if you do want the extra 3D glits, installing the proprietory NVidia drivers is as trivial as this:
As root:
- rpm -ivh http://rpm.livna.org/livna-release-7.rpm
- rpm --import /etc/pki/rpm-gpg/RPM-GPG-KEY-livna
- yum install kmod-nvidia
I almost agree with you. On point 3 you are implicitly assuming that the users card is new enough to be support by the latest driver release. Not all are. Livna also support various older versions for these cards, and so if you need to use these you will need a different kmod.
Chris
Alan Cox wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
When the users' interests aren't the same as the project's interests, you have to wonder why there are any users - or how long there will be any.
On Sun, 2007-10-28 at 18:42 +0000, Alan Cox wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
Are you saying the driver is illegal full stop ? If so in what way ?
Chris
On Sun, 28 Oct 2007 19:06:21 +0000 Chris Jones jonesc@hep.phy.cam.ac.uk wrote:
Are you saying the driver is illegal full stop ?
It may be.
If so in what way ?
It adds a binary blob to the kernel.
I just found a reference that explains this rather well.
QUOTE: The majority of core Linux contributors believe (based on legal advice) that it's impossible to produce a loadable Linux kernel module without it being a derivative of the kernel code to some extent, and therefore having to fall under the GPL due to the GPL's restrictions on derivative works. It's not a situation that's ever been tested in court, and clearly not everyone agrees - however, it's plausible enough that there's basically no chance of code being added to the core kernel for no purpose other than facilitating a set of non-GPLed kernel modules. END OF QUOTE
You can find this (and further discussion) here: http://blogs.sun.com/ahl/entry/dtrace_knockoffs
on 10/28/2007 3:04 PM, Les Mikesell wrote:
Alan Cox wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
When the users' interests aren't the same as the project's interests, you have to wonder why there are any users - or how long there will be any.
The Linux users should get of their collective lazy a$$es and setup their systems to suit themselves if they want to use non FOSS packages.
Since the current subject is Nvidia, a Windows install, for example, comes with a 'it works' Nvidia video driver. Just like Linux does. *Not* the fancy one from Nvidia that does all of the fancy stuff. The 'Stupid Windows User' must go to the Nvidia site, find the proper driver for his video card and install it to get the fancy video stuff.
So what you are saying or implying here? That 'Stupid Windows Users' know more about their systems and how to set them up than 'Brilliant Linux Users' do? ;-)
On Sun, 2007-10-28 at 17:25 +0000, Jonathan Underwood wrote:
On 28/10/2007, Matthew Saltzman mjs@clemson.edu wrote:
In the past, I have found that Nvidia has been pretty good compared with other manufacturers (especially ATI) about their proprietary drivers. Now that ATI is opening their drivers, I will be seriously considering them in the future. I'm stuck with the laptop for a while, though 8^(.
At this point in time I have purged my life of all nvidia cards in favour of intel which have OSS drivers contributed by Intel. They may not be high end gaming cards, but they work damn well under linux, and so I'd recommend them highly if that is an option for you.
I'll look at them too for my next machines, but for my home machines, I like the 3D capability, and as I said, for the laptop, I don't get a choice. (The T61 has an Intel video option, and I wish the powers that be had chosen it, but they didn't, and I didn't know it was going to turn out to be such a hassle.)
On Sun, 28 Oct 2007 15:22:51 -0400 Matthew Saltzman mjs@CLEMSON.EDU wrote:
I'll look at them too for my next machines, but for my home machines, I like the 3D capability,
I'm not really a gamer (at all) but I did take a look at Alien Arena the other day and somewhat to my surprise it worked perfectly with my Intel video card.
Brutal Chess also works fine and looks lovely.
Les Mikesell wrote:
Claude Jones wrote:
On Sun October 28 2007, Les Mikesell wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
I think any but the most recent members of this list have noted your position on that subject, Les --- it would have been hard to miss '>
And yet, people still try to run fedora on machines with nvidia and seem surprised by the problems...
I was one who bought a new computer and until today could not tell you had a particular nvitia video card. Today I see the whole computer is full of things made by nvidia. And I was one that tried to run Fedora on it and had the problems and I was surprised. Nobody told me it was bad. I might not have been able to even tell the motherboard had nvidia. The motherboard got rave reviews from Windows users. Why not Fedora?
David Boles wrote:
on 10/28/2007 3:04 PM, Les Mikesell wrote:
Alan Cox wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
When the users' interests aren't the same as the project's interests, you have to wonder why there are any users - or how long there will be any.
The Linux users should get of their collective lazy a$$es and setup their systems to suit themselves if they want to use non FOSS packages.
And why is that? Especially if they believe that vendors should be able to distribute their drivers in whatever form they want?
Since the current subject is Nvidia, a Windows install, for example, comes with a 'it works' Nvidia video driver. Just like Linux does. *Not* the fancy one from Nvidia that does all of the fancy stuff. The 'Stupid Windows User' must go to the Nvidia site, find the proper driver for his video card and install it to get the fancy video stuff.
So what you are saying or implying here? That 'Stupid Windows Users' know more about their systems and how to set them up than 'Brilliant Linux Users' do? ;-)
My reply was in response to an 'it doesn't work' post. And with Windows there is one packaging method and a small number of interface variations to pick from. Linux has presented a near infinite variety with not much indication of getting it right yet.
Alan Cox wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
For God's sake why is something that you can d/l from their web site in the USA be illegal in the USA? Why does it not meet the Fedora policy? Where can I read the policy?
If your stating the intent of the policy is to diminish the availability of Fedora to new users then the policy needs to change.
Chris Jones wrote:
Firstly, Fedora will work out of the box with nvidia cards using the free/OSS drivers. They may not yet properly support 3D, but they do work and give you a graphical interface.
At that point, if you do want the extra 3D glits, installing the proprietory NVidia drivers is as trivial as this:
As root:
- rpm -ivh http://rpm.livna.org/livna-release-7.rpm
- rpm --import /etc/pki/rpm-gpg/RPM-GPG-KEY-livna
- yum install kmod-nvidia
I almost agree with you. On point 3 you are implicitly assuming that the users card is new enough to be support by the latest driver release. Not all are. Livna also support various older versions for these cards, and so if you need to use these you will need a different kmod.
Chris
But if you have a fix for the new cards at least your that much further along. And we have no data on how well old hardware will work with new drivers.
Frank Cox wrote:
On Sun, 28 Oct 2007 19:06:21 +0000 Chris Jones jonesc@hep.phy.cam.ac.uk wrote:
Are you saying the driver is illegal full stop ?
It may be.
If so in what way ?
It adds a binary blob to the kernel.
I just found a reference that explains this rather well.
QUOTE: The majority of core Linux contributors believe (based on legal advice) that it's impossible to produce a loadable Linux kernel module without it being a derivative of the kernel code to some extent, and therefore having to fall under the GPL due to the GPL's restrictions on derivative works. It's not a situation that's ever been tested in court, and clearly not everyone agrees - however, it's plausible enough that there's basically no chance of code being added to the core kernel for no purpose other than facilitating a set of non-GPLed kernel modules. END OF QUOTE
You can find this (and further discussion) here: http://blogs.sun.com/ahl/entry/dtrace_knockoffs
Nothing you say involves legal problems. They are just rules kernel people like you to use. But there is no rule that says I can't compile the kernel myself. So it all sounds silly to me.
David Boles wrote:
on 10/28/2007 3:04 PM, Les Mikesell wrote:
Alan Cox wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
When the users' interests aren't the same as the project's interests, you have to wonder why there are any users - or how long there will be any.
The Linux users should get of their collective lazy a$$es and setup their systems to suit themselves if they want to use non FOSS packages.
Since the current subject is Nvidia, a Windows install, for example, comes with a 'it works' Nvidia video driver. Just like Linux does. *Not* the fancy one from Nvidia that does all of the fancy stuff. The 'Stupid Windows User' must go to the Nvidia site, find the proper driver for his video card and install it to get the fancy video stuff.
So what you are saying or implying here? That 'Stupid Windows Users' know more about their systems and how to set them up than 'Brilliant Linux Users' do? ;-)
The Windows user has a screen with a pointer on it. He can use that screen forever and be happy. If he is at home and wants to run Flight Simulator he needs to get the better driver and the instruction on flight sim include how to get it.
Fedora users discover a broken screen sometimes without a pointer and that is how it works. Stupid since there are alternatives.
On Sun, 2007-28-10 at 13:58 -0600, Karl Larsen wrote:
For God's sake why is something that you can d/l from their web site
in the USA be illegal in the USA? Why does it not meet the Fedora policy? Where can I read the policy?
Try the front page of the Fedora project. For more detail, see http://fedoraproject.org/wiki/Overview
If your stating the intent of the policy is to diminish the
availability of Fedora to new users then the policy needs to change.
The fact that, earlier this month, the Fedora marketing list was considering adapting the slogan "Freedom is a feature" should tell you something. Fedora's policy is to ship nothing that isn't free software -- and drivers whose code is proprietary don't meet the definition, even if they are free for the download.
This policy is such a cornerstone of Fedora that it is highly unlikely to change.
Other distributions have different policies. If using the non-free drivers provided by NVidia is a priority for you, then, with all respect, perhaps you should look into them instead of using Fedora.
On Sun, 28 Oct 2007 14:06:52 -0600 Karl Larsen k5di@zianet.com wrote:
Nothing you say involves legal problems. They are just rules kernel people like you to use.
That's completely wrong, but don't let that stop you, Karl.
But there is no rule that says I can't compile the kernel myself.
You can do what you wish with the kernel that you're using on your own personal machine. But we're talking about distribution here, not compiling it yourself. So... where did you get the modifications that you're compiling?
From someone who's distributing them, perhaps?
So it all sounds silly to me.
Again, you're completely wrong.
But if you have a fix for the new cards at least your that much further along. And we have no data on how well old hardware will work with new drivers.
Easy, just use the readme from the nvidia site to find out which driver supports your card
http://us.download.nvidia.com/XFree86/Linux-x86/100.14.19/README/index.html
(appendix A)
and then choose the correct driver from livna for your card.
Chris
For God's sake why is something that you can d/l from their web site
in the USA be illegal in the USA? Why does it not meet the Fedora
For the same reason that if I take Star Wars and put it up on my web site George Lucas will consider it illegal. Copyright law. The fact its up on my website doesn't make it legal
policy? Where can I read the policy?
I'd try the Fedora web site.
If your stating the intent of the policy is to diminish the
availability of Fedora to new users then the policy needs to change.
The policy is designed to - Comply with the law - Produce a free software based distribution - Work in the -long term- interests of the free software and Linux world.
Alan
Bruce Byfield wrote:
On Sun, 2007-28-10 at 13:58 -0600, Karl Larsen wrote:
For God's sake why is something that you can d/l from their web site
in the USA be illegal in the USA? Why does it not meet the Fedora policy? Where can I read the policy?
Try the front page of the Fedora project. For more detail, see http://fedoraproject.org/wiki/Overview
If your stating the intent of the policy is to diminish the
availability of Fedora to new users then the policy needs to change.
The fact that, earlier this month, the Fedora marketing list was considering adapting the slogan "Freedom is a feature" should tell you something. Fedora's policy is to ship nothing that isn't free software -- and drivers whose code is proprietary don't meet the definition, even if they are free for the download.
This policy is such a cornerstone of Fedora that it is highly unlikely to change.
Other distributions have different policies. If using the non-free drivers provided by NVidia is a priority for you, then, with all respect, perhaps you should look into them instead of using Fedora.
You didn't need to mention that I have a choice. I can change to Windows which due to it's cost includes what a person with my computer needs to work. Fedora does not and plans if your accurate to make it even harder for a user to use your free product. That sounds counter productive to me.
Karl Larsen wrote:
I was one who bought a new computer and until today could not tell
you had a particular nvitia video card.
If I was going to buy a new computer and wanted to run Linux on it I would at least google for "<new computer> Linux" to see what problems I was likely to meet.
On Sun, 2007-28-10 at 14:45 -0600, Karl Larsen wrote:
You didn't need to mention that I have a choice. I can change to
Windows which due to it's cost includes what a person with my computer needs to work. Fedora does not and plans if your accurate to make it even harder for a user to use your free product.
Why "even harder"? The policy isn't changing. It's the same one that Fedora has always used.
That sounds counter productive to me.
That depends on whether you value software freedom or not. If you do, then Fedora is doing exactly what it should be doing, and a few inconveniences are an acceptable price to pay for maintaining your political and philosophical opinion.
If you don't -- if all you want is an operating system that doesn't cost you anything and operates as efficiently as possible -- then, yeah, I suppose it must sound counter-productive. But, in that case, your priorities are greatly out of line with Fedora's.
Timothy Murphy wrote:
Karl Larsen wrote:
I was one who bought a new computer and until today could not tell
you had a particular nvitia video card.
If I was going to buy a new computer and wanted to run Linux on it I would at least google for "<new computer> Linux" to see what problems I was likely to meet.
I did. But it was the motherboard since the computer was a kit from a supplier. what I got was many web pages that were thrilled with how well it worked. And another thing you forget. At that time I was not aware that nvidia was bad. I had never even heard of it.
Karl Larsen wrote:
I was one who bought a new computer and until today could not tell
you had a particular nvitia video card.
If I was going to buy a new computer and wanted to run Linux on it I would at least google for "<new computer> Linux" to see what problems I was likely to meet.
I did. But it was the motherboard since the computer was a kit from a supplier. what I got was many web pages that were thrilled with how well it worked. And another thing you forget. At that time I was not aware that nvidia was bad. I had never even heard of it.
Nvidia isn't 'bad' in any normal sense. They just have not made source code available for their drivers available under GPL terms as some small fraction of users would like to demand. It it the fact the Linux will not provide a stable interface (so a working binary module continues to work) and the fedora project's policy not to cooperate with anyone with different terms than their own that make it difficult to use the combination. I'd guess that it's easier to change the OS than a motherboard chip...
Nvidia isn't 'bad' in any normal sense. They just have not made source code available for their drivers available under GPL terms as some small fraction of users would like to demand.
Please see the COPYING file and the GPL licence. If you are creating a derivative work of the Linux kernel as a lot of us think the Nvidia drivers for Linux are then you are required to abide by the licence, in exactly the same way as you are required to obey other licences.
Alan
Les Mikesell wrote:
Karl Larsen wrote:
I was one who bought a new computer and until today could not tell
you had a particular nvitia video card.
If I was going to buy a new computer and wanted to run Linux on it I would at least google for "<new computer> Linux" to see what problems I was likely to meet.
I did. But it was the motherboard since the computer was a kit from a supplier. what I got was many web pages that were thrilled with how well it worked. And another thing you forget. At that time I was not aware that nvidia was bad. I had never even heard of it.
Nvidia isn't 'bad' in any normal sense. They just have not made source code available for their drivers available under GPL terms as some small fraction of users would like to demand. It it the fact the Linux will not provide a stable interface (so a working binary module continues to work) and the fedora project's policy not to cooperate with anyone with different terms than their own that make it difficult to use the combination. I'd guess that it's easier to change the OS than a motherboard chip...
Thanks Les. I realize that wanting Fedora to work even on computers that use nvidia video cards is a lost cause with Fedora. I put "linux,nvidia" in Goggle and I see all the other Linux brands are providing the binary in their kit.
on 10/28/2007 3:55 PM, Les Mikesell wrote:
David Boles wrote:
on 10/28/2007 3:04 PM, Les Mikesell wrote:
Alan Cox wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
When the users' interests aren't the same as the project's interests, you have to wonder why there are any users - or how long there will be any.
The Linux users should get of their collective lazy a$$es and setup their systems to suit themselves if they want to use non FOSS packages.
And why is that? Especially if they believe that vendors should be able to distribute their drivers in whatever form they want?
Since the current subject is Nvidia, a Windows install, for example, comes with a 'it works' Nvidia video driver. Just like Linux does. *Not* the fancy one from Nvidia that does all of the fancy stuff. The 'Stupid Windows User' must go to the Nvidia site, find the proper driver for his video card and install it to get the fancy video stuff.
So what you are saying or implying here? That 'Stupid Windows Users' know more about their systems and how to set them up than 'Brilliant Linux Users' do? ;-)
My reply was in response to an 'it doesn't work' post. And with Windows there is one packaging method and a small number of interface variations to pick from. Linux has presented a near infinite variety with not much indication of getting it right yet.
So you would expect Nvidia to spend a ton of money for a very few users/buyers? And they, the users, can not even decide just how that they should provide the package? Rpm? Deb? Tar.gz. Zip?
I have an idea for you. Cut a deal with Nvidia and you do this for them. All of this work. On your watch. And on your bank account. And then give the product away. Tell me when you startup will you?? ;-)
BTW - I don't use Nvidia's drivers. But I have installed them to see just how difficult this is. It took longer to open a can of soda than the actual compile and install took. And, GEE, when I updated my kernel I had to do it all over again. Man!!! What a difficult thing to do!!!
Get real here. The bottom line is that no matter what you want or say here... It ain't going to happen.
on 10/28/2007 4:11 PM, Karl Larsen wrote:
David Boles wrote:
on 10/28/2007 3:04 PM, Les Mikesell wrote:
Alan Cox wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
When the users' interests aren't the same as the project's interests, you have to wonder why there are any users - or how long there will be any.
The Linux users should get of their collective lazy a$$es and setup their systems to suit themselves if they want to use non FOSS packages.
Since the current subject is Nvidia, a Windows install, for example, comes with a 'it works' Nvidia video driver. Just like Linux does. *Not* the fancy one from Nvidia that does all of the fancy stuff. The 'Stupid Windows User' must go to the Nvidia site, find the proper driver for his video card and install it to get the fancy video stuff.
So what you are saying or implying here? That 'Stupid Windows Users' know more about their systems and how to set them up than 'Brilliant Linux Users' do? ;-)
The Windows user has a screen with a pointer on it. He can use that
screen forever and be happy. If he is at home and wants to run Flight Simulator he needs to get the better driver and the instruction on flight sim include how to get it.
Fedora users discover a broken screen sometimes without a pointer
and that is how it works. Stupid since there are alternatives.
I didn't say that you should *not* use the Nvidia drivers Karl. Only that you should be prepared to do the work needed to do that. And that you should understand just *why* Fedora can *not* do this. And just why the people on the 'third party sites' might be a little behind your needs. Thinks like they might have real jobs. They are doing it because they want to do it. And for no money. Kinda' hard to be upset with someone who is giving you their time and efforts for free. That's all.
On Sun, 28 Oct 2007 17:53:23 -0400 David Boles dgboles@gmail.com wrote:
So you would expect Nvidia to spend a ton of money for a very few users/buyers?
If they want to be in the Linux market and if that's what it takes. Yes.
And they, the users, can not even decide just how that they should provide the package? Rpm? Deb? Tar.gz. Zip?
All of the above would do quite nicely. That's only three formats and each one has the same actual content.
I have an idea for you. Cut a deal with Nvidia and you do this for them. All of this work. On your watch. And on your bank account. And then give the product away. Tell me when you startup will you?? ;-)
Nvidia gives their video cards away for free now? I hadn't heard that....
David Boles wrote:
on 10/28/2007 4:11 PM, Karl Larsen wrote:
David Boles wrote:
on 10/28/2007 3:04 PM, Les Mikesell wrote:
Alan Cox wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
When the users' interests aren't the same as the project's interests, you have to wonder why there are any users - or how long there will be any.
The Linux users should get of their collective lazy a$$es and setup their systems to suit themselves if they want to use non FOSS packages.
Since the current subject is Nvidia, a Windows install, for example, comes with a 'it works' Nvidia video driver. Just like Linux does. *Not* the fancy one from Nvidia that does all of the fancy stuff. The 'Stupid Windows User' must go to the Nvidia site, find the proper driver for his video card and install it to get the fancy video stuff.
So what you are saying or implying here? That 'Stupid Windows Users' know more about their systems and how to set them up than 'Brilliant Linux Users' do? ;-)
The Windows user has a screen with a pointer on it. He can use that
screen forever and be happy. If he is at home and wants to run Flight Simulator he needs to get the better driver and the instruction on flight sim include how to get it.
Fedora users discover a broken screen sometimes without a pointer
and that is how it works. Stupid since there are alternatives.
I didn't say that you should *not* use the Nvidia drivers Karl. Only that you should be prepared to do the work needed to do that. And that you should understand just *why* Fedora can *not* do this. And just why the people on the 'third party sites' might be a little behind your needs. Thinks like they might have real jobs. They are doing it because they want to do it. And for no money. Kinda' hard to be upset with someone who is giving you their time and efforts for free. That's all.
Well after all this "why we can't do a dam thing" let's consider how to help a new user of Fedora x who stumbles onto this list and is asking for help with the usual nvidia problems. We need someone to get those people headed in the right direction. I will be willing to do that. As I looked today nvidia is a big company making a lot of money. My computer is full of there stuff. Since Fedora x will just work terrible with nvidia I think we must be ready to help.
Alan Cox wrote:
Nvidia isn't 'bad' in any normal sense. They just have not made source code available for their drivers available under GPL terms as some small fraction of users would like to demand.
Please see the COPYING file and the GPL licence. If you are creating a derivative work of the Linux kernel as a lot of us think the Nvidia drivers for Linux are then you are required to abide by the licence, in exactly the same way as you are required to obey other licences.
It's very imaginative to consider a device driver for a device that clearly works under other operating systems to be a derivative of one of those OS's. And the position would be a little more believable if accompanied by some court decisions that permitted interfaces needed for normal operation to be copyrighted. But I don't think you really want the latter...
In any case the result would just be that nvidia owners won't be able to run Linux. And the real question is, why do you want that - or even to make it more difficult than necessary for those users?
On Sun, 28 Oct 2007 17:58:30 -0500 Les Mikesell lesmikesell@gmail.com wrote:
In any case the result would just be that nvidia owners won't be able to run Linux. And the real question is, why do you want that - or even to make it more difficult than necessary for those users?
Assuming that you really do want an answer to your question, here it is:
To further the goals of Free Software, and to encourage outfits like Nvidia to see the light, as it were, and provide Free Software drivers for their hardware like so many other hardware manufacturers already do.
On Sun, 2007-10-28 at 15:59 -0600, Karl Larsen wrote:
Les Mikesell wrote:
Karl Larsen wrote:
I was one who bought a new computer and until today could not tell
you had a particular nvitia video card.
If I was going to buy a new computer and wanted to run Linux on it I would at least google for "<new computer> Linux" to see what problems I was likely to meet.
I did. But it was the motherboard since the computer was a kit from a supplier. what I got was many web pages that were thrilled with how well it worked. And another thing you forget. At that time I was not aware that nvidia was bad. I had never even heard of it.
Nvidia isn't 'bad' in any normal sense. They just have not made source code available for their drivers available under GPL terms as some small fraction of users would like to demand. It it the fact the Linux will not provide a stable interface (so a working binary module continues to work) and the fedora project's policy not to cooperate with anyone with different terms than their own that make it difficult to use the combination. I'd guess that it's easier to change the OS than a motherboard chip...
Thanks Les. I realize that wanting Fedora to work even on computers
that use nvidia video cards is a lost cause with Fedora. I put "linux,nvidia" in Goggle and I see all the other Linux brands are providing the binary in their kit.
---- there's a lot of mis-information here
xorg includes an nv module (for nvidia video) which is open source and part of xorg/fedora packaging. This driver does 2D.
nVidia corporation also provides a binary only driver that is designed for specific kernel builds and since it is binary only, patent encumbered, restricted license, not open source, fedora packaging will not include that. Livna and other repositories will typically package it with a kernel-mod which sort of makes it painless for the user (the user must install the yum repository package for livna but that's about it).
nVidia binary modules seem to work reasonably well but it is not necessary to use them at all.
Any other Linux distribution that packages patent encumbered, restricted license, binary only modules into their base packaging and distributed with other GPL license software would likely be violating the GPL license, run afoul of patent restrictions in the US and would certainly be distributing software that they have absolutely no possible way to certify as safe and/or maintainable.
Craig
Frank Cox wrote:
On Sun, 28 Oct 2007 17:53:23 -0400 David Boles dgboles@gmail.com wrote:
Here is what nvidia makes: NVIDIA Corporation (NVIDIA) is engaged in the provision of programmable graphics processor technologies. The Company has four product-line operating segments: graphics processing units (GPUs), media and communications processors (MCPs), Handheld GPU Business, and Consumer
It has a Market Cap of 1888 Billion shares at $35.00 each. It trades on the Nasdaq board. I happen to have 200 shares. I like the stock.
On Sun, 28 Oct 2007 17:05:22 -0600 Karl Larsen k5di@zianet.com wrote:
Here is what nvidia makes:
Did you have an actual point to make or a comment or a question, or are you just trying to waste more of everyone's bandwidth?
Hint: Everyone else here can type nvidia.com into their web browser too.
Thus, Karl Larsen at Sun Oct 28 22:43:12 2007 inscribed: [snip]
Well after all this "why we can't do a dam thing" let's consider how to help a new user of Fedora x who stumbles onto this list and is asking for help with the usual nvidia problems. We need someone to get those people headed in the right direction. I will be willing to do that. As I looked today nvidia is a big company making a lot of money. My computer is full of there stuff. Since Fedora x will just work terrible with nvidia I think we must be ready to help.
As far as I can tell, those users does receive help, and sometimes a nudge in the direction of graphics cards that don't need the closed-source, proprietary drivers. (Like Intel graphics cards or now recently, ATI.)
It is also not rocket-surgery to head over to nvidia.com and click through to the driver download site. It even works in elinks. Out-of-the-box, Fedora tends to work rather well with nVidia graphics cards.
It is admirable that you are volounteering to help these new users though.
On Sun, 28 Oct 2007 23:11:04 +0000 Anders Karlsson anders@trudheim.co.uk wrote:
It is admirable that you are volounteering to help these new users though.
I lean toward using a different adjective. "Scary" comes to mind. Perhaps "unfortunate".
on 10/28/2007 6:43 PM, Karl Larsen wrote:
David Boles wrote:
on 10/28/2007 4:11 PM, Karl Larsen wrote:
David Boles wrote:
on 10/28/2007 3:04 PM, Les Mikesell wrote:
Alan Cox wrote:
> But note that there is nothing remotely illegal about the nvidia driver, > nor would there be any problem with the fedora project distributing it > themselves if they had any reason to care about the user experience. > > Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
When the users' interests aren't the same as the project's interests, you have to wonder why there are any users - or how long there will be any.
The Linux users should get of their collective lazy a$$es and setup their systems to suit themselves if they want to use non FOSS packages.
Since the current subject is Nvidia, a Windows install, for example, comes with a 'it works' Nvidia video driver. Just like Linux does. *Not* the fancy one from Nvidia that does all of the fancy stuff. The 'Stupid Windows User' must go to the Nvidia site, find the proper driver for his video card and install it to get the fancy video stuff.
So what you are saying or implying here? That 'Stupid Windows Users' know more about their systems and how to set them up than 'Brilliant Linux Users' do? ;-)
The Windows user has a screen with a pointer on it. He can use that
screen forever and be happy. If he is at home and wants to run Flight Simulator he needs to get the better driver and the instruction on flight sim include how to get it.
Fedora users discover a broken screen sometimes without a pointer
and that is how it works. Stupid since there are alternatives.
I didn't say that you should *not* use the Nvidia drivers Karl. Only that you should be prepared to do the work needed to do that. And that you should understand just *why* Fedora can *not* do this. And just why the people on the 'third party sites' might be a little behind your needs. Thinks like they might have real jobs. They are doing it because they want to do it. And for no money. Kinda' hard to be upset with someone who is giving you their time and efforts for free. That's all.
Well after all this "why we can't do a dam thing" let's consider how
to help a new user of Fedora x who stumbles onto this list and is asking for help with the usual nvidia problems. We need someone to get those people headed in the right direction. I will be willing to do that. As I looked today nvidia is a big company making a lot of money. My computer is full of there stuff. Since Fedora x will just work terrible with nvidia I think we must be ready to help.
There you go again Karl. A good idea. Really. But you do need a website or a blog for this information. When it comes to an information source this list is okay but try to search for past questions and answers. Which is why you see the same questions asked again and again and answered in the same or similar ways over and over.
What I find truly amazing is the fact that some (most) Newbies will wipe out a perfectly good working Mac OS X system, or a perfectly good working Windows system, and install a completely strange Linux system without doing any preparations. Like reading stuff and learning a little about it *before* they do this. Dumb things like will my computer still work after I install Linux? Can I still listen to music? Can I still watch videos? Can I still play my favorite game? Can I get on The Internet? It goes on.
As for the repeating questions and problems that I mentioned? Fedora 8 is about two weeks away. I won't take the time or bandwidth to list them here but I already know what the subjects of the posts for next several weeks will be before they happen. Care to know? Go back and check the Fedora 7 release posts. Or the Fedora Core 6 posts. Or the Fedora Core 5 posts. And on and on.
On Sun, 2007-28-10 at 17:58 -0500, Les Mikesell wrote:
It's very imaginative to consider a device driver for a device that clearly works under other operating systems to be a derivative of one of those OS's.
You're being disingenuous. In this case, a derivative work is not a casual term, but a concept defined by the second version of the GNU General Public License.
And the position would be a little more believable if accompanied by some court decisions that permitted interfaces needed for normal operation to be copyrighted. But I don't think you really want the latter...
The whole point is that the language of GPL2 is ambiguous enough that the status of Nvidia proprietary drivers is uncertain. Few lawyers with any subject matter expertise are going to recommend that a distribution ship the drivers when their status is uncertain, even if they are willing to. A court case might actually determine the status.
However, regardless of the legal status, the drivers aren't released under a free license, so they aren't going to be released by any distribution concerned with software freedom.
It's very imaginative to consider a device driver for a device that clearly works under other operating systems to be a derivative of one of
Fine - just load the windows version on your Linux box.. that should be .. fun ...
those OS's. And the position would be a little more believable if accompanied by some court decisions that permitted interfaces needed for normal operation to be copyrighted. But I don't think you really want the latter...
Nothing to do with interfaces, only whether something is a derivative work. Based on discussions with lawyers a lot of us think it is. Since Nvidia still ship it I would assume their lawyer thinks it isnt.
In any case the result would just be that nvidia owners won't be able to run Linux. And the real question is, why do you want that - or even to make it more difficult than necessary for those users?
A very strange argument. Applied to movies you'd argue we should all be able to pirate them as charging makes it difficult for the users ?
There is a project to do open source 3D for Nvidia cards - look up Nouveau. There is also one working on drivers for the newer ATI hardware, although they at least now have the documentation.
For the future I'll leave you with this thought as to why I don't worry about Nvidia.
- Everyone agrees graphics will get integrated onto the CPU die - AMD/ATI have a fast x86 CPU - Intel have a fast x86 CPU - Nvidia don't
Alan
Alan Cox wrote:
In any case the result would just be that nvidia owners won't be able to run Linux. And the real question is, why do you want that - or even to make it more difficult than necessary for those users?
A very strange argument. Applied to movies you'd argue we should all be able to pirate them as charging makes it difficult for the users ?
There's no piracy involved here as nvidia gives the drivers away and permits redistribution. It is the Linux side that makes their product difficult or impossible to use, with the obvious effect that the majority of users keep running Microsoft products and the rest of us have to put up with the resulting virus-laden machines.
On Sun, 28 Oct 2007 19:38:35 -0500 Les Mikesell lesmikesell@gmail.com wrote:
There's no piracy involved here as nvidia gives the drivers away and permits redistribution.
That depends entirely on your point of view.
It can be argued that Nvidia is pirating the work of the Linux kernel developers when creating their closed-source driver for Linux systems.
on 10/28/2007 8:38 PM, Les Mikesell wrote:
Alan Cox wrote:
A very strange argument. Applied to movies you'd argue we should all be able to pirate them as charging makes it difficult for the users ?
There's no piracy involved here as nvidia gives the drivers away and permits redistribution. It is the Linux side that makes their product difficult or impossible to use, with the obvious effect that the majority of users keep running Microsoft products and the rest of us have to put up with the resulting virus-laden machines.
And *all*, not just Fedora but *all*, of the Linux distributions are 'doing this' just because they *know* how much this upsets you. ;-)
On 10/28/07, Karl Larsen k5di@zianet.com wrote:
You might have a nvidia video card on your motherboard. There are
two choices here. Try to use the nvidia or turn it off and plug in your old known video card. Today I wish I had done the latter because using nvidia with f7 is a pain.
I'm starting to wonder if Karl is not sent here by Microsoft to perpetuate misinformation. I mean, really. This is about the 20th thread that starts of with "I'm killing myself doing xyz" and then we later find out there's nothing wrong or that there are already published instructions.
I've owned Nvidia cards for years. All types of cards. Fedora always finds the cards, the default nv driver does install and run adequately. Installing the Nvidia proprietary driver is quite simple and easy for anybody that is willing to read a paragraph on the nvidia website. And if you are not up for reading a full paragraph, there have been RPMs for the nvidia drivers for years.
In conclusion, I'm not falling for this bait and don't think anybody else should either.
On Sunday 28 October 2007 12:28:56 pm Karl Larsen wrote:
Ralf Corsepius wrote:
On Sun, 2007-10-28 at 12:00 -0400, Carroll Grigsby wrote:
According to the Soyo site, your motherboard has an onboard Prosavage graphics chip (http://www.soyousa.com/products/proddesc.php?t=d&id=292), and uses a VIA chipset. Nothing there about nVidia, so I assume that the nVidia card is an add-on.
/sbin/lspci probably will tell.
Ralf
Thanks Ralf, you are right again. Here is a whole list of nvidia
hardware:
[root@k5di ~]# lspci 00:00.0 RAM memory: nVidia Corporation C51 Host Bridge (rev a2) 00:00.1 RAM memory: nVidia Corporation C51 Memory Controller 0 (rev a2) 00:00.2 RAM memory: nVidia Corporation C51 Memory Controller 1 (rev a2) 00:00.3 RAM memory: nVidia Corporation C51 Memory Controller 5 (rev a2) 00:00.4 RAM memory: nVidia Corporation C51 Memory Controller 4 (rev a2) 00:00.5 RAM memory: nVidia Corporation C51 Host Bridge (rev a2) 00:00.6 RAM memory: nVidia Corporation C51 Memory Controller 3 (rev a2) 00:00.7 RAM memory: nVidia Corporation C51 Memory Controller 2 (rev a2) 00:03.0 PCI bridge: nVidia Corporation C51 PCI Express Bridge (rev a1) 00:04.0 PCI bridge: nVidia Corporation C51 PCI Express Bridge (rev a1) 00:05.0 VGA compatible controller: nVidia Corporation C51G [GeForce 6100] (rev a2) 00:09.0 RAM memory: nVidia Corporation MCP51 Host Bridge (rev a2) 00:0a.0 ISA bridge: nVidia Corporation MCP51 LPC Bridge (rev a3) 00:0a.1 SMBus: nVidia Corporation MCP51 SMBus (rev a3) 00:0a.2 RAM memory: nVidia Corporation MCP51 Memory Controller 0 (rev a3) 00:0b.0 USB Controller: nVidia Corporation MCP51 USB Controller (rev a3) 00:0b.1 USB Controller: nVidia Corporation MCP51 USB Controller (rev a3) 00:0d.0 IDE interface: nVidia Corporation MCP51 IDE (rev a1) 00:0e.0 IDE interface: nVidia Corporation MCP51 Serial ATA Controller (rev a1) 00:10.0 PCI bridge: nVidia Corporation MCP51 PCI Bridge (rev a2) 00:10.2 Multimedia audio controller: nVidia Corporation MCP51 AC97 Audio Controller (rev a2) 00:14.0 Bridge: nVidia Corporation MCP51 Ethernet Controller (rev a3) 00:18.0 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] HyperTransport Technology Configuration 00:18.1 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Address Map 00:18.2 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] DRAM Controller 00:18.3 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Miscellaneous Control [root@k5di ~]#
And the video controller is a nVidia Corporation C51G [GeForce 6100]
(rev a2) what ever that might mean. Also the entire computer is full of nvidia things and the only one causing trouble is the video.
Karl: First, whatever Soyo board that you have, it certainly isn't the model number that you cited above.
Second, unlike video cards, onboard chips _usually_ rely on system RAM rather than onboard RAM. If that is the case here, you have to reserve some memory space for the use of the chip by using a mem=XXXX command at boot time. (xxxx is the difference between the installed RAM and that required by the video card). I've never dealt with one of these onboard critters, so I'll leave it to you to determine the exact form of the mem command. In particular, be careful about expressing the amount of memory -- I'm not certain whether it's in kilobytes, megabyte or whatever.
-- cmg
--- Les Mikesell lesmikesell@gmail.com wrote:
Alan Cox wrote:
In any case the result would just be that nvidia
owners won't be able to
run Linux. And the real question is, why do you
want that - or even to
make it more difficult than necessary for those
users?
A very strange argument. Applied to movies you'd
argue we should all be
able to pirate them as charging makes it difficult
for the users ?
For movies, pirate them, pirate is not a good word here. Users should be able to make backup copy of their owned dvd's. Buying another dvd to replace a scratched or lost dvd makes no sense since they are expen$ive. The end users have to suffer from market prices for movies since the cost is high and pirate movies work and for alot le$$.
There's no piracy involved here as nvidia gives the drivers away and permits redistribution.
Gives the drivers away, but not the source which is more important for the principles of Free Software. The binary blob is free, but not the source thereby impling a closed source project.
It is the Linux side that makes their product difficult or impossible to use, with the obvious effect that the majority of users keep running Microsoft products and the rest of us have to put up with the resulting virus-laden machines.
-- Les Mikesell lesmikesell@gmail.com
--
You can use the drivers with Fedora or any other linux for that matter, but they are not to be bothered if the machine freezes, burns or does not work as it should. Several users that have nvidia cards or other ones that need special drivers are aware of the facts and still run the drivers and have no problems and therefore do not complain. Fedora will never take responsibility for closed source software/drivers not working correctly. You are at your own risk.
It would be great though that one day nvidia would release their code and users around the world can improve upon it and release it so that then Fedora would incorporate the new code and get 3D rendering and the stuff that many users want, till that day comes, users will have to use that software at their own risk and with no guarantees.
Mr Alan Cox has a very interesting explanation bout open vs closed source. To read more about open-source development vs closed source development by Mr. Cox go here" http://blogs.cnet.com/8301-13505_1-9803919-16.html
I have several nvidia cards on my computers and the nv driver works nicely with no need for 3D deskops, or games. Basic computing is enough for me, I tried the binary driver and the tv out did not work :( so might as well give up and use the driver that just works and move on.
Regards,
Antonio
__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
Frank Cox wrote:
On Sun, 28 Oct 2007 19:38:35 -0500 Les Mikesell lesmikesell@gmail.com wrote:
There's no piracy involved here as nvidia gives the drivers away and permits redistribution.
That depends entirely on your point of view.
It can be argued that Nvidia is pirating the work of the Linux kernel developers when creating their closed-source driver for Linux systems.
Before you make that argument, could you explain why you want to make it? That is, why do Linux developers hate the users who would benefit from this combination and want to drive them into Microsoft's arms?
Then, even if you do make it, you'll have to establish why this isn't ordinary fair use of an interface required to interoperate with the kernel.
Antonio Olivares wrote:
Fedora will never take responsibility for closed source software/drivers not working correctly. You are at your own risk.
Wait - are you saying that they _do_ take responsibility for the open source they distribute? Who should I call the next time my firewire drives aren't recognized or an update kernel won't boot? How soon will someone be over to fix it?
Les Mikesell wrote:
Antonio Olivares wrote:
Fedora will never take responsibility for closed source software/drivers not working correctly. You are at your own risk.
Wait - are you saying that they _do_ take responsibility for the open source they distribute?
To the extend that any Free software project does.
Who should I call the next time my firewire
drives aren't recognized or an update kernel won't boot? How soon will someone be over to fix it?
If you want the ability to call someone or get a guarantee to fix problems, get a commercial support contract. RHEL is one option.
Rahul
on 10/28/2007 9:56 PM, Les Mikesell wrote:
Antonio Olivares wrote:
Fedora will never take responsibility for closed source software/drivers not working correctly. You are at your own risk.
Wait - are you saying that they _do_ take responsibility for the open source they distribute? Who should I call the next time my firewire drives aren't recognized or an update kernel won't boot? How soon will someone be over to fix it?
As soon as you pay for the software and the service call.
You get what you pay for ya' know. ;-)
David Boles wrote:
on 10/28/2007 9:56 PM, Les Mikesell wrote:
Antonio Olivares wrote:
Fedora will never take responsibility for closed source software/drivers not working correctly. You are at your own risk.
Wait - are you saying that they _do_ take responsibility for the open source they distribute? Who should I call the next time my firewire drives aren't recognized or an update kernel won't boot? How soon will someone be over to fix it?
As soon as you pay for the software and the service call.
You get what you pay for ya' know. ;-)
And thus there is no difference at all in this respect between the third party binary and an open source component. Except that the people providing the binary do have a reason to care if it works.
On Sun, 28 Oct 2007 21:29:09 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Except that the people providing the binary do have a reason to care if it works.
Until they decide that it's time to sell you another card and discontinue their binary blob for the model that you already have.
On Sun, 28 Oct 2007 20:51:03 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Before you make that argument, could you explain why you want to make it?
Could it be because it's true?
That is, why do Linux developers hate the users who would benefit from this combination and want to drive them into Microsoft's arms?
That's a different question entirely. I don't think Linux developers hate users. I'd imagine that most have an attitude similar to mine: If someone wants to run Linux they are welcome to do so. If they choose to run Microsoft software instead, that's their decision. Their actions don't bother me either way. I will help someone out when they require assistance but I refuse to help those who won't help themselves (i.e. move away from Mother Microsoft).
Again, that's just me but I don't think I'm alone in my thinking here.
Then, even if you do make it, you'll have to establish why this isn't ordinary fair use of an interface required to interoperate with the kernel.
Because it's contrary to the license under which the Linux kernel is distributed. I don't think it can be made much clearer than that, frankly.
Frank Cox wrote:
On Sun, 28 Oct 2007 21:29:09 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Except that the people providing the binary do have a reason to care if it works.
Until they decide that it's time to sell you another card and discontinue their binary blob for the model that you already have.
You mean like when a developer of an opensource project decides he doesn't want or can't continue to support the project and it closes down for lack of others picking up the mantle?
Of course nvidia doesn't do what you are suggesting. They have their legacy channel as in:
Latest Legacy GPU version (1.0-71xx series): 71.86.01 Latest Legacy GPU version (1.0-96xx series): 96.43.01
On Sun, 28 Oct 2007 21:29:09 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Except that the people providing the binary do have a reason to care if it works.
Considering that NVidia has offered the absolute minimal to keep GNU/Linux users installing their cards, and will do nothing to help the Nouveau team reverse engineer the drivers, this is really a pretty funny statement.
Ed Greshko wrote:
Frank Cox wrote:
On Sun, 28 Oct 2007 21:29:09 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Except that the people providing the binary do have a reason to care if it works.
Until they decide that it's time to sell you another card and discontinue their binary blob for the model that you already have.
You mean like when a developer of an opensource project decides he doesn't want or can't continue to support the project and it closes down for lack of others picking up the mantle?
The difference is that, there is a opportunity for others to get involved which is not the case for non-free software. Most popular Free software projects frequently are supported by multiple groups even commercially and when one goes away, another steps up to fill in the gap as it has happened quite often and there is also the possibility that you hire others or get in-house people to maintain it. non-free software does give that level of control to end users.
This is apart from the legal issues involved in combining non-free modules and the Linux kernel for a distribution.
Rahul
on 10/28/2007 10:29 PM, Les Mikesell wrote:
David Boles wrote:
on 10/28/2007 9:56 PM, Les Mikesell wrote:
Antonio Olivares wrote:
Fedora will never take responsibility for closed source software/drivers not working correctly. You are at your own risk.
Wait - are you saying that they _do_ take responsibility for the open source they distribute? Who should I call the next time my firewire drives aren't recognized or an update kernel won't boot? How soon will someone be over to fix it?
As soon as you pay for the software and the service call.
You get what you pay for ya' know. ;-)
And thus there is no difference at all in this respect between the third party binary and an open source component. Except that the people providing the binary do have a reason to care if it works.
Not really. Nvidia releases buggy, doesn't work well drivers from time to time.
If you find a Linux distribution, a free one, that is different from what your attitude describes will you post the name here?
All that I can tell you is that *I* am happy. As must be many others since there are only a few complaints.
What is your latest disaster this time anyway? I missed that part of this thread.
Frank Cox wrote:
Then, even if you do make it, you'll have to establish why this isn't ordinary fair use of an interface required to interoperate with the kernel.
Because it's contrary to the license under which the Linux kernel is distributed. I don't think it can be made much clearer than that, frankly.
It's not at all clear why the kernel license should control someone else's work.
On Sun, 28 Oct 2007 21:56:36 -0500 Les Mikesell lesmikesell@gmail.com wrote:
It's not at all clear why the kernel license should control someone else's work.
Because their project is built "into" the kernel developer's work.
If I rent a house to you, don't you think I should be entitled to evict you if you decide to start ripping out the walls and remodelling the place?
On Sun, 2007-10-28 at 13:43 -0400, Claude Jones wrote:
On Sun October 28 2007, Les Mikesell wrote:
But note that there is nothing remotely illegal about the nvidia driver, nor would there be any problem with the fedora project distributing it themselves if they had any reason to care about the user experience.
I think any but the most recent members of this list have noted your position on that subject, Les --- it would have been hard to miss '>
If nVidia would allow it's driver to be distributed freely, that would be a good thing. The Croquet project will not run as it should using the yum-able versions of the nVidia drivers. Les has experienced the same problem with Croquet, as we both have the 5200 card and only installing the nVidia direct package will make it run as it is supposed to. We both had openGL problems out the wazoo. The nVidia package fixes that.
I have no clue exactly what is in the nVidia supplied package as opposed to the ones from Livna or Freshrpms, but something (proprietary?) apparently is missing and the application demands something not found in the rpm versions. Back to the age-old "Right Thing" imperative, I have to use what works. No apologies for that.
You know me, Claude ...I bitched for months trying to figure out what was wrong, before finally getting Croquet to run. When the solution appeared, I was floored for wasting those months. While I'm grateful that nVidia supports linux and offer a driver package, I hope that nVidia will open their code, so this kinda stuff doesn't happen to anyone else. I seriously doubt that they would lose one bit of their sales. Ric
--- Les Mikesell lesmikesell@gmail.com wrote:
Frank Cox wrote:
Then, even if you do make it, you'll have to
establish why this isn't
ordinary fair use of an interface required to
interoperate with the kernel.
Because it's contrary to the license under which
the Linux kernel is
distributed. I don't think it can be made much
clearer than that, frankly.
It's not at all clear why the kernel license should control someone else's work.
-- Les Mikesell lesmikesell@gmail.com
--
Remember Les, the kernel is released under the GPL, and as such it is restricted to the conditions provided under it. There are other licenses which might not restrict this, but the kernel and all the drivers that make their way into it fall under this umbrella. It is a very touchy issue and there's not much that we can do to change it:( .
For me and others, the nv driver works fine :) and it is open source, comes with Fedora and other distros.
Regards,
Antonio
__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
Frank Cox wrote:
Except that the people providing the binary do have a reason to care if it works.
Until they decide that it's time to sell you another card and discontinue their binary blob for the model that you already have.
In a reasonable OS, a binary blob that worked today would work again tomorrow.
--- Les Mikesell lesmikesell@gmail.com wrote:
Antonio Olivares wrote:
Fedora will never take responsibility for closed source software/drivers
not
working correctly. You are at your own risk.
Wait - are you saying that they _do_ take responsibility for the open source they distribute?
I am not implying that, but since you conclude that on your own, the upstream developers and the fedora developers work hand in hand to fix code and release fixes for them. But for nvidia code and drivers they are not to take any blame for the success/failure of the binary drivers. That's why there is the "use at your risk" part in the statement.
Who should I call the next time my firewire drives aren't recognized or an update kernel won't boot?
You can do a number of things that might/might not get answered (1) File Bug Reports (2) File RFE's (3) complain, complain, complain, and hope that someone listens and tries to help (4) wait patiently till a fix makes its way via updates
How soon will someone be over to fix it?
Depends on the issues and the severity of the bugs. Developers work hard and try to release updates to fix problems.
-- Les Mikesell lesmikesell@gmail.com
--
Regards,
Antonio
__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
On Sun, 28 Oct 2007 22:11:29 -0500 Les Mikesell lesmikesell@gmail.com wrote:
In a reasonable OS, a binary blob that worked today would work again tomorrow.
I think the word you're looking for is moribund.
On Sun, 2007-10-28 at 22:11 -0500, Les Mikesell wrote:
Frank Cox wrote:
Except that the people providing the binary do have a reason to care if it works.
Until they decide that it's time to sell you another card and discontinue their binary blob for the model that you already have.
In a reasonable OS, a binary blob that worked today would work again tomorrow.
---- you mean like a stable environment such as RHEL / CentOS ?
Craig
On Sun, 28 Oct 2007 20:14:13 -0700 (PDT) Antonio Olivares olivares14031@yahoo.com wrote:
You can do a number of things that might/might not get answered (1) File Bug Reports (2) File RFE's (3) complain, complain, complain, and hope that someone listens and tries to help (4) wait patiently till a fix makes its way via updates
You missed two options: (5) Pay someone to fix it for you (6) Attempt to fix it yourself and probably learn a great deal in the process of doing so, regardless of whether you meet with ultimate success or not.
I think option 6 is the most valuable option over the long term.
David Boles wrote:
If you find a Linux distribution, a free one, that is different from what your attitude describes will you post the name here?
Most of my machines run Centos, where the interface doesn't change every week.
What is your latest disaster this time anyway? I missed that part of this thread.
My only recent ones have been the kernels that wouldn't boot on IBM MPT scsi controllers. I gave up on fedora on any machine with nvidia or firewire long ago.
Frank Cox wrote:
In a reasonable OS, a binary blob that worked today would work again tomorrow.
I think the word you're looking for is moribund.
I was thinking more of "correctly designed". What's the point of changing it after you get it right?
Antonio Olivares wrote:
It's not at all clear why the kernel license should control someone else's work.
Remember Les, the kernel is released under the GPL, and as such it is restricted to the conditions provided under it. There are other licenses which might not restrict this, but the kernel and all the drivers that make their way into it fall under this umbrella. It is a very touchy issue and there's not much that we can do to change it:( .
Linus was widely quoted as saying that binary modules were permitted in the early years. Personally I don't believe it would have become popular had that not been so.
And touchy or not, someone would have to prove that a driver was a derived work before the GPL could prevent distribution under other terms.
Rahul Sundaram wrote:
Ed Greshko wrote:
Frank Cox wrote:
On Sun, 28 Oct 2007 21:29:09 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Except that the people providing the binary do have a reason to care if it works.
Until they decide that it's time to sell you another card and discontinue their binary blob for the model that you already have.
You mean like when a developer of an opensource project decides he doesn't want or can't continue to support the project and it closes down for lack of others picking up the mantle?
The difference is that, there is a opportunity for others to get involved which is not the case for non-free software. Most popular Free software projects frequently are supported by multiple groups even commercially and when one goes away, another steps up to fill in the gap as it has happened quite often and there is also the possibility that you hire others or get in-house people to maintain it. non-free software does give that level of control to end users.
That is a "difference" but one has to understand that first, the nvidia drivers are free...just not open and second, there is a corporation behind the nvidia drivers and it is certainly possible that the corporation values its reputation.
In both commercial and opensource software there are no guarantees that a particular piece of software will be supported or continue to be developed.
This is apart from the legal issues involved in combining non-free modules and the Linux kernel for a distribution.
I must have missed that part of the discussion. I didn't notice it was part of the discussion that nvidia drivers or ATI drivers be made part of a distro.
Ed Greshko wrote:
That is a "difference" but one has to understand that first, the nvidia drivers are free...just not open and second, there is a corporation behind the nvidia drivers and it is certainly possible that the corporation values its reputation.
You don't have to rely on a single vendor with Free software.
In both commercial and opensource software there are no guarantees that a particular piece of software will be supported or continue to be developed.
The distinction between commercial and FOSS is a false one. You mean proprietary or non-free software. There is a lot of commercial FOSS applications and products available today.
Neither offers guarantees but Free software offers more choice and flexibility if a vendor decides to cut off support for a product.
This is apart from the legal issues involved in combining non-free modules and the Linux kernel for a distribution.
I must have missed that part of the discussion. I didn't notice it was part of the discussion that nvidia drivers or ATI drivers be made part of a distro.
Read back. That indeed is a major thrust of the discussion.
Rahul
Rahul Sundaram wrote:
The distinction between commercial and FOSS is a false one. You mean proprietary or non-free software. There is a lot of commercial FOSS applications and products available today.
I didn't mean to imply that... But, it is also unfair to automatically label free (as in no cost) proprietary software as "bad".
Neither offers guarantees but Free software offers more choice and flexibility if a vendor decides to cut off support for a product.
And a good vendor won't do that.
I wish you wouldn't use the term "Free Software" as it is open to interpretation of what one means by "Free".
This is apart from the legal issues involved in combining non-free modules and the Linux kernel for a distribution.
I must have missed that part of the discussion. I didn't notice it was part of the discussion that nvidia drivers or ATI drivers be made part of a distro.
Read back. That indeed is a major thrust of the discussion.
Too many tangents.
Ed Greshko wrote:
Rahul Sundaram wrote:
The distinction between commercial and FOSS is a false one. You mean proprietary or non-free software. There is a lot of commercial FOSS applications and products available today.
I didn't mean to imply that... But, it is also unfair to automatically label free (as in no cost) proprietary software as "bad".
That depends on perspective. Fedora's perspective is to support Free and open source software exclusively. That doesn't require being judgmental about non-free software.
Neither offers guarantees but Free software offers more choice and flexibility if a vendor decides to cut off support for a product.
And a good vendor won't do that.
Many vendors do drop support quickly (which makes sense since the focus is on volume) and some go out of business. Free software projects can route around such damage.
Rahul
On Sun, 28 Oct 2007 22:36:04 -0500 Les Mikesell lesmikesell@gmail.com wrote:
I was thinking more of "correctly designed". What's the point of changing it after you get it right?
Technological advances, research, new ideas....
Why aren't you running Z-80 CP/M with dual floppies and a green-screen monitor?
On 10/28/07, Karl Larsen k5di@zianet.com wrote:
David Boles wrote:
on 10/28/2007 4:11 PM, Karl Larsen wrote:
David Boles wrote:
on 10/28/2007 3:04 PM, Les Mikesell wrote:
Alan Cox wrote:
> But note that there is nothing remotely illegal about the nvidia driver, > nor would there be any problem with the fedora project distributing it > themselves if they had any reason to care about the user experience. > > Actually the legal state of the Nvidia driver is a matter of debate. Even if it was legal it would not meet the Fedora policy, nor be in the interests of the project.
When the users' interests aren't the same as the project's interests, you have to wonder why there are any users - or how long there will be any.
The Linux users should get of their collective lazy a$$es and setup their systems to suit themselves if they want to use non FOSS packages.
Since the current subject is Nvidia, a Windows install, for example, comes with a 'it works' Nvidia video driver. Just like Linux does. *Not* the fancy one from Nvidia that does all of the fancy stuff. The 'Stupid Windows User' must go to the Nvidia site, find the proper driver for his video card and install it to get the fancy video stuff.
So what you are saying or implying here? That 'Stupid Windows Users' know more about their systems and how to set them up than 'Brilliant Linux Users' do? ;-)
The Windows user has a screen with a pointer on it. He can use that
screen forever and be happy. If he is at home and wants to run Flight Simulator he needs to get the better driver and the instruction on flight sim include how to get it.
Fedora users discover a broken screen sometimes without a pointer
and that is how it works. Stupid since there are alternatives.
I didn't say that you should *not* use the Nvidia drivers Karl. Only that you should be prepared to do the work needed to do that. And that you should understand just *why* Fedora can *not* do this. And just why the people on the 'third party sites' might be a little behind your needs. Thinks like they might have real jobs. They are doing it because they want to do it. And for no money. Kinda' hard to be upset with someone who is giving you their time and efforts for free. That's all.
Well after all this "why we can't do a dam thing" let's consider how
to help a new user of Fedora x who stumbles onto this list and is asking for help with the usual nvidia problems. We need someone to get those people headed in the right direction. I will be willing to do that. As I looked today nvidia is a big company making a lot of money. My computer is full of there stuff. Since Fedora x will just work terrible with nvidia I think we must be ready to help.
since you want to help to really resolve the problem the nouveau project is waiting for your patches: http://nouveau.freedesktop.org/wiki/
--
Karl F. Larsen, AKA K5DI Linux User #450462 http://counter.li.org.
-- fedora-list mailing list fedora-list@redhat.com To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
Rahul Sundaram wrote:
This is apart from the legal issues involved in combining non-free modules and the Linux kernel for a distribution.
I must have missed that part of the discussion. I didn't notice it was part of the discussion that nvidia drivers or ATI drivers be made part of a distro.
Read back. That indeed is a major thrust of the discussion.
There are many ways the situation could be improved for the user. The best I would hope for would be the inclusion of a configuration for a yum repository containing the drivers currently maintained in the livna repo. If that's too much to ask, then a link to the repo from the fedora project site would be better than nothing. Actually including the driver or maintaining a stable interface would be too much to expect here.
On Sun, 28 Oct 2007 23:55:10 -0500 Les Mikesell lesmikesell@gmail.com wrote:
There are many ways the situation could be improved for the user.
The proper way to improve the situation for the user is for Nvidia to abide by the GPL and provide their drivers in the normal and correct manner.
Anything else is an un-necessary work-around or accommodation that is not and should not be required by the Fedora developers or anyone else.
Frank Cox wrote:
I was thinking more of "correctly designed". What's the point of changing it after you get it right?
Technological advances, research, new ideas....
OK, so once a decade or so there might be a worthwhile change.
Why aren't you running Z-80 CP/M with dual floppies and a green-screen monitor?
I did have the source for those drivers... Wasn't your point that having source was supposed to make it usable forever?
Les Mikesell wrote:
There are many ways the situation could be improved for the user. The best I would hope for would be the inclusion of a configuration for a yum repository containing the drivers currently maintained in the livna repo. If that's too much to ask, then a link to the repo from the fedora project site would be better than nothing. Actually including the driver or maintaining a stable interface would be too much to expect here.
Again, Fedora will only include Free and open source software. Supporting non-free software is any manner is a explicit non-objective.
http://fedoraproject.org/wiki/Objectives
Rahul
Rahul Sundaram wrote:
Ed Greshko wrote:
Rahul Sundaram wrote:
The distinction between commercial and FOSS is a false one. You mean proprietary or non-free software. There is a lot of commercial FOSS applications and products available today.
I didn't mean to imply that... But, it is also unfair to automatically label free (as in no cost) proprietary software as "bad".
That depends on perspective. Fedora's perspective is to support Free and open source software exclusively. That doesn't require being judgmental about non-free software.
Well...nvidia drivers are "free" as no cost. So, I'm confused.
Neither offers guarantees but Free software offers more choice and flexibility if a vendor decides to cut off support for a product.
And a good vendor won't do that.
Many vendors do drop support quickly (which makes sense since the focus is on volume) and some go out of business. Free software projects can route around such damage.
Sure, vendors do drop support. Not sure it is all about "volume" since innovation comes into play. Yet, generally, and in the case of nvidia, when the hardware is no longer produced/sold the drivers continue to be available.
You say "Free software projects" but I think you mean "OpenSource" or maybe your definition of "Free" is different than mine.
While an opensouce project "can" route around such damage it doesn't mean they "will".
On Sun, 28 Oct 2007 23:01:07 -0600 Frank Cox theatre@sasktel.net wrote:
by the Fedora developers or anyone else.
*ahem*
_OF_ the Fedora developers or anyone else.
....
Sorry about that, Chief!
On Mon, 29 Oct 2007 00:01:59 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Why aren't you running Z-80 CP/M with dual floppies and a green-screen monitor?
I did have the source for those drivers... Wasn't your point that having source was supposed to make it usable forever?
Yeah, pretty much. I'm sure you could still use if you wanted to; there are a lot of drivers for dot matrix printers that you can use with your Linux setup, for example. Software emulators work rather well, too. (I use Vice and DOSEMU on a fairly regular basis.)
On Mon, 29 Oct 2007 13:02:15 +0800 Ed Greshko Ed.Greshko@greshko.com wrote:
Well...nvidia drivers are "free" as no cost. So, I'm confused.
Here you go:
http://www.fsf.org/licensing/essays/free-sw.html
I hope this is helpful. It covers the meaning of Free Software quite clearly.
Ed Greshko wrote:
Well...nvidia drivers are "free" as no cost. So, I'm confused.
Look up the definition of non-free software.
You say "Free software projects" but I think you mean "OpenSource" or maybe your definition of "Free" is different than mine.
Free as in freedom. Not free as in gratis.
While an opensouce project "can" route around such damage it doesn't mean they "will".
The opportunity is clearly more. That certainly counts as an advantage especially for distributions who have to fix bugs or include enhancements.
Rahul
Rahul Sundaram wrote:
Ed Greshko wrote:
Well...nvidia drivers are "free" as no cost. So, I'm confused.
Look up the definition of non-free software.
Who's definition? When a term can be ambiguous it is of little value to me.
You say "Free software projects" but I think you mean "OpenSource" or maybe your definition of "Free" is different than mine.
Free as in freedom. Not free as in gratis.
While an opensouce project "can" route around such damage it doesn't mean they "will".
The opportunity is clearly more. That certainly counts as an advantage especially for distributions who have to fix bugs or include enhancements.
The opportunity may be there...but it doesn't mean (at least to me) that it counts as a slam dunk advantage. Sorry, I tend not to be overly religious on these sort of issues.
Ed Greshko wrote:
Rahul Sundaram wrote:
Ed Greshko wrote:
Well...nvidia drivers are "free" as no cost. So, I'm confused.
Look up the definition of non-free software.
Who's definition? When a term can be ambiguous it is of little value to me.
Use your favorite search engine and search for "Free software" and "non-free". Read about it.
The opportunity may be there...but it doesn't mean (at least to me) that it counts as a slam dunk advantage. Sorry, I tend not to be overly religious on these sort of issues.
It counts as an advantage for distributions like Fedora which distribute Free software which is a focus on long-term advantage. Whether it matters personally to you or upto to you to decide.
Rahul
Rahul Sundaram wrote:
Les Mikesell wrote:
There are many ways the situation could be improved for the user. The best I would hope for would be the inclusion of a configuration for a yum repository containing the drivers currently maintained in the livna repo. If that's too much to ask, then a link to the repo from the fedora project site would be better than nothing. Actually including the driver or maintaining a stable interface would be too much to expect here.
Again, Fedora will only include Free and open source software. Supporting non-free software is any manner is a explicit non-objective.
We all know that the real objective of fedora is to get a lot of users to do the beta testing on what will be the next RHEL after which they quickly stop supporting those tested versions and move on. Maybe they get enough testers without having to be bothered making it convenient for nvidia, vmware, etc. users.
Les Mikesell wrote:
Again, Fedora will only include Free and open source software. Supporting non-free software is any manner is a explicit non-objective.
We all know that the real objective of fedora is to get a lot of users to do the beta testing on what will be the next RHEL after which they quickly stop supporting those tested versions and move on. Maybe they get enough testers without having to be bothered making it convenient for nvidia, vmware, etc. users.
Promoting Free and open source software is a well recognized objective of Fedora. You might choose to ignore it and insert your own perspective. Doesn't change the fact that Fedora isn't interested in supporting proprietary software.
Rahul
Frank Cox wrote:
There are many ways the situation could be improved for the user.
The proper way to improve the situation for the user is for Nvidia to abide by the GPL and provide their drivers in the normal and correct manner.
They have claimed that they can't because parts are licensed from other parties.
Anything else is an un-necessary work-around or accommodation that is not and should not be required by the Fedora developers or anyone else.
Do you not recognize that there are other existing licenses on technology that users need?
Rahul Sundaram wrote:
Les Mikesell wrote:
Again, Fedora will only include Free and open source software. Supporting non-free software is any manner is a explicit non-objective.
We all know that the real objective of fedora is to get a lot of users to do the beta testing on what will be the next RHEL after which they quickly stop supporting those tested versions and move on. Maybe they get enough testers without having to be bothered making it convenient for nvidia, vmware, etc. users.
Promoting Free and open source software is a well recognized objective of Fedora. You might choose to ignore it and insert your own perspective. Doesn't change the fact that Fedora isn't interested in supporting proprietary software.
There's a big difference in 'supporting' proprietary software and shooting the feet out from under your own users that depend on it.
Les Mikesell wrote:
Promoting Free and open source software is a well recognized objective of Fedora. You might choose to ignore it and insert your own perspective. Doesn't change the fact that Fedora isn't interested in supporting proprietary software.
There's a big difference in 'supporting' proprietary software and shooting the feet out from under your own users that depend on it.
Fedora Project is very clear about the fact that it includes and supports Free and open source software. Not proprietary software in any form. It is upto users to see if the design meet their needs and choose appropriately. There is nothing done explicitly to antagonize users using non-free software nor is there anything done to accommodate them.
Rahul
On Mon, 29 Oct 2007 00:36:47 -0500 Les Mikesell lesmikesell@gmail.com wrote:
There's a big difference in 'supporting' proprietary software and shooting the feet out from under your own users that depend on it.
If you depend on proprietary software, then Linux is not for you. Microsoft and/or Apple products are a much better fit for your needs.
On Mon, 29 Oct 2007 00:28:48 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Do you not recognize that there are other existing licenses on technology that users need?
There are all kinds of licenses. If your requirements include "proprietary, closed-source and immutable software", then Linux is not for you.
I'm sorry if this is the first time you have heard this.
Thus, Les Mikesell at Mon Oct 29 03:11:29 2007 inscribed:
Frank Cox wrote:
Except that the people providing the binary do have a reason to care if it works.
Until they decide that it's time to sell you another card and discontinue their binary blob for the model that you already have.
In a reasonable OS, a binary blob that worked today would work again tomorrow.
Les,
If you have such a beef about this particularly part about the argument, I suggest you take it onto the Linux Kernel mailing list. I would suggest you search the archives for it first though, as it has been debated extensively already and all the arguments you can come up with most likely have been replied to. It'll save traffic on this list.
Just an observation...
Rahul Sundaram wrote:
Who's definition? When a term can be ambiguous it is of little value to me.
Use your favorite search engine and search for "Free software" and "non-free". Read about it.
OK... By that I take it there is no one universally accepted definition of those terms. Otherwise, there would be no need for all the different flavors of the "free" licenses. GPLv2, GPLv3, FreeBSD, etc...
And, it would be easy to define it here.
The opportunity may be there...but it doesn't mean (at least to me) that it counts as a slam dunk advantage. Sorry, I tend not to be overly religious on these sort of issues.
It counts as an advantage for distributions like Fedora which distribute Free software which is a focus on long-term advantage. Whether it matters personally to you or upto to you to decide.
On Sun, 2007-10-28 at 20:05 -0700, Craig White wrote:
On Sun, 2007-10-28 at 22:37 -0400, David Boles wrote:
What is your latest disaster this time anyway? I missed that part of this thread.
Les' mission is to save us from the evils of GPL license.
Not so, no where did Les say that, Craig. I know Les ...he has a bunch of coding chops under his belt and he's a helluva decent human being. The point is that it needn't be anathema if the only way to get a device working correctly is to use a driver from the manufacturer. Sure, pang on them until the cows come home to open up their code. I believe in that. One day everyone will see * * The Light * *. Guys like Alan Cox could tell them a thing or three or ten about proper interfacing to the kernel.
Meanwhile, getting stuff to work is a consideration for Joe Lunch de Bucket, Les, me and a bunch of others. And, we deal with what we got as we can. :) Ric
Frank Cox wrote:
On Mon, 29 Oct 2007 00:36:47 -0500 Les Mikesell lesmikesell@gmail.com wrote:
There's a big difference in 'supporting' proprietary software and shooting the feet out from under your own users that depend on it.
If you depend on proprietary software, then Linux is not for you. Microsoft and/or Apple products are a much better fit for your needs.
Never did care that much for blanket statements.
If I depend on say "Oracle" and since Oracle is proprietary I should not run it on Linux?
On Mon, 29 Oct 2007 15:24:35 +0800 Ed Greshko Ed.Greshko@greshko.com wrote:
OK... By that I take it there is no one universally accepted definition of those terms. Otherwise, there would be no need for all the different flavors of the "free" licenses. GPLv2, GPLv3, FreeBSD, etc...
Each one of those licenses serves a particular purpose or need. (Whether each one is actually required is a different question.)
With regard to Free Software in particular, in the modern computer industry, the generally accepted definition of Free Software (notice the capitalization) is the definition that is provided and promoted by the Free Software Foundation. Which is what I sent you to see.
It's a kind of an "industry standard definition", for lack of a better term.
Many industrial trades, for example, use otherwise common terms that mean something "special" in terms of their specific industry. Free Software is a good example of an otherwise common term that has a special meaning in the computer industry.
On Mon, 29 Oct 2007 15:29:46 +0800 Ed Greshko Ed.Greshko@greshko.com wrote:
If you depend on proprietary software, then Linux is not for you. Microsoft and/or Apple products are a much better fit for your needs.
Never did care that much for blanket statements.
If I depend on say "Oracle" and since Oracle is proprietary I should not run it on Linux?
You are taking a general statement and applying it to a very specific circumstance.
In the particular case cited, Linux may indeed be the answer you seek. However, in general terms, the solutions that folks like Les are looking for can be most easily found in places other than Linux. Microsoft and Apple are the two largest examples of such places.
On Sun, 2007-10-28 at 22:40 -0600, Frank Cox wrote:
On Sun, 28 Oct 2007 22:36:04 -0500 Les Mikesell lesmikesell@gmail.com wrote:
I was thinking more of "correctly designed". What's the point of changing it after you get it right?
Technological advances, research, new ideas....
Why aren't you running Z-80 CP/M with dual floppies and a green-screen monitor?
I wish I had mine back! <grins> I had three Altos, three IMSAI's, one Osburn luggable, Televideo 802, 803 networked to a 1603 running MP/M II, one Kaypro, Jeeeezzz, I ferget the rest of the pile. Nothing better than a green-screen monitor hooked to a behemoth Sperry-Unisys 5000/90 and watch the display speeding along a line at the time at 9600 baud. Wordstar! Super-Calc! Streaming tape! That's what tar really is for. You betcha. I really miss the old iron. :) Ric
On Sun, 2007-10-28 at 22:37 -0400, David Boles wrote:
on 10/28/2007 10:29 PM, Les Mikesell wrote:
David Boles wrote:
on 10/28/2007 9:56 PM, Les Mikesell wrote:
Antonio Olivares wrote:
Fedora will never take responsibility for closed source software/drivers not working correctly. You are at your own risk.
Wait - are you saying that they _do_ take responsibility for the open source they distribute? Who should I call the next time my firewire drives aren't recognized or an update kernel won't boot? How soon will someone be over to fix it?
As soon as you pay for the software and the service call.
You get what you pay for ya' know. ;-)
And thus there is no difference at all in this respect between the third party binary and an open source component. Except that the people providing the binary do have a reason to care if it works.
Not really. Nvidia releases buggy, doesn't work well drivers from time to time.
In my experience, the nv drivers at this moment are buggy and don't work well.
https://bugzilla.redhat.com/show_bug.cgi?id=249367
The proprietary drivers don't work perfectly with that card either, but they do actually work.
What was your point?
If you find a Linux distribution, a free one, that is different from what your attitude describes will you post the name here?
All that I can tell you is that *I* am happy. As must be many others since there are only a few complaints.
What is your latest disaster this time anyway? I missed that part of this thread.
On Mon, 29 Oct 2007 03:45:29 -0400 Ric Moore wayward4now@gmail.com wrote:
I really miss the old iron.
I don't have much experience with other CP/M emulators, but the CP/M mode of the Commodore 128 is a great deal of fun to play with through the Vice emulator.
You can get it here: http://viceteam.org/
On Mon, 29 Oct 2007 03:57:24 -0400 Matthew Saltzman mjs@CLEMSON.EDU wrote:
In my experience, the nv drivers at this moment are buggy and don't work well.
https://bugzilla.redhat.com/show_bug.cgi?id=249367
The proprietary drivers don't work perfectly with that card either, but they do actually work.
What was your point?
Sounds to me very much like "the point" is that one should stay far far away from Nvidia, circumstances being what they are at the moment.
Many of us have been telling you (the general "you", that is) that very thing for some time now.
Anders Karlsson wrote:
If you have such a beef about this particularly part about the argument, I suggest you take it onto the Linux Kernel mailing list. I would suggest you search the archives for it first though, as it has been debated extensively already and all the arguments you can come up with most likely have been replied to. It'll save traffic on this list.
Actually, to quote Linus on LKML, in discussion of whether or not binary modules are "derived" or not, specifically related to Nvidia:
"I think the NVidia people can probably reasonably honestly say that the code they ported had _no_ Linux origin." [1]
Yes, I'm quite aware that many people will simply argue that it doesn't matter what Linus thinks as he is not the sole copyright holder in the kernel, and so on and so forth. Save it -- that's already at LKML too. Just wanted to point out that the LKML archives are not as clear on this point as you make out.
Cheers, Raman
On Sun, 28 Oct 2007 21:56:36 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Frank Cox wrote:
Then, even if you do make it, you'll have to establish why this isn't ordinary fair use of an interface required to interoperate with the kernel.
Because it's contrary to the license under which the Linux kernel is distributed. I don't think it can be made much clearer than that, frankly.
It's not at all clear why the kernel license should control someone else's work.
If the work is derived it isn't someone elses work, any more than if you add a scene to a movie and redistribute it. Thats the fundamental question - is the Nvidia driver a derived work. Some people believe no, some believe yes, nobody "knows" in the US legal sense because nobody has been to court to find out - and I suspect neither side considers it worth finding out.
Alan
There's no piracy involved here as nvidia gives the drivers away and permits redistribution. It is the Linux side that makes their product
If I give away copies of someone elses movies and permit redistribution then thats still piracy. The question is one of derived works and what the derived work boundary is.
Ed Greshko wrote:
Rahul Sundaram wrote:
Who's definition? When a term can be ambiguous it is of little value to me.
Use your favorite search engine and search for "Free software" and "non-free". Read about it.
OK... By that I take it there is no one universally accepted definition of those terms. Otherwise, there would be no need for all the different flavors of the "free" licenses. GPLv2, GPLv3, FreeBSD, etc...
And, it would be easy to define it here.
Sure there is. You just got to read the definition that comes up. All the licenses you quote above fit into the definition.
Rahul
on 10/28/2007 11:32 PM, Les Mikesell wrote:
David Boles wrote:
If you find a Linux distribution, a free one, that is different from what your attitude describes will you post the name here?
Most of my machines run Centos, where the interface doesn't change every week.
What is your latest disaster this time anyway? I missed that part of this thread.
My only recent ones have been the kernels that wouldn't boot on IBM MPT scsi controllers. I gave up on fedora on any machine with nvidia or firewire long ago.
Is that "IBM MPT scsi controllers" only? That does sound like a possible bug. Did you write a ticket on that?
Strange. Which version (release) of Fedora? Core 6? Fedora 7? I ask because I have an Nvidia graphics card and the firewire on my machine.
I was using Nvidia graphics cards when I first tried Linux at RedHat v5.2. I think it was v5.2. It came with a "Linux for Dummies" book. I built my current machine around the time of Fedora 7. I, personally, do not have anything firewire but my son's camcorder works when attached. It did in Fedora 7 and still does in Fedora 8.
Thus, Raman Gupta at Mon Oct 29 08:57:07 2007 inscribed:
Anders Karlsson wrote:
[snip: suggestion to query the license part on LKML]
Actually, to quote Linus on LKML, in discussion of whether or not binary modules are "derived" or not, specifically related to Nvidia:
"I think the NVidia people can probably reasonably honestly say that the code they ported had _no_ Linux origin." [1]
Yes, I'm quite aware that many people will simply argue that it doesn't matter what Linus thinks as he is not the sole copyright holder in the kernel, and so on and so forth. Save it -- that's already at LKML too. Just wanted to point out that the LKML archives are not as clear on this point as you make out.
This is true, and this is not a simple matter, even for legally trained people. I simply think that while there is a question open on the legality of certain drivers and packages, arguing about their licenses and that Fedora should incorporate them no matter what is counter-productive to the aims of *this* list.
The legal issue stems from the kernel side and the use of the module interface to said kernel. This is not a distribution issue, this is an issue related to the kernel, and as such, should in all honesty be debated on the mailing list devoted to the kernel.
YMMV
In the particular case cited, Linux may indeed be the answer you seek. However, in general terms, the solutions that folks like Les are looking for can be most easily found in places other than Linux. Microsoft and Apple are the two largest examples of such places.
There are also other linux distros that don't take the same line as fedora. Ubuntu being the obvious example, that does make it much easier for users to use things like the nvidia driver.
Chris
Carroll Grigsby wrote:
On Sunday 28 October 2007 12:28:56 pm Karl Larsen wrote:
Ralf Corsepius wrote:
On Sun, 2007-10-28 at 12:00 -0400, Carroll Grigsby wrote:
According to the Soyo site, your motherboard has an onboard Prosavage graphics chip (http://www.soyousa.com/products/proddesc.php?t=d&id=292), and uses a VIA chipset. Nothing there about nVidia, so I assume that the nVidia card is an add-on.
/sbin/lspci probably will tell.
Ralf
Thanks Ralf, you are right again. Here is a whole list of nvidia
hardware:
[root@k5di ~]# lspci 00:00.0 RAM memory: nVidia Corporation C51 Host Bridge (rev a2) 00:00.1 RAM memory: nVidia Corporation C51 Memory Controller 0 (rev a2) 00:00.2 RAM memory: nVidia Corporation C51 Memory Controller 1 (rev a2) 00:00.3 RAM memory: nVidia Corporation C51 Memory Controller 5 (rev a2) 00:00.4 RAM memory: nVidia Corporation C51 Memory Controller 4 (rev a2) 00:00.5 RAM memory: nVidia Corporation C51 Host Bridge (rev a2) 00:00.6 RAM memory: nVidia Corporation C51 Memory Controller 3 (rev a2) 00:00.7 RAM memory: nVidia Corporation C51 Memory Controller 2 (rev a2) 00:03.0 PCI bridge: nVidia Corporation C51 PCI Express Bridge (rev a1) 00:04.0 PCI bridge: nVidia Corporation C51 PCI Express Bridge (rev a1) 00:05.0 VGA compatible controller: nVidia Corporation C51G [GeForce 6100] (rev a2) 00:09.0 RAM memory: nVidia Corporation MCP51 Host Bridge (rev a2) 00:0a.0 ISA bridge: nVidia Corporation MCP51 LPC Bridge (rev a3) 00:0a.1 SMBus: nVidia Corporation MCP51 SMBus (rev a3) 00:0a.2 RAM memory: nVidia Corporation MCP51 Memory Controller 0 (rev a3) 00:0b.0 USB Controller: nVidia Corporation MCP51 USB Controller (rev a3) 00:0b.1 USB Controller: nVidia Corporation MCP51 USB Controller (rev a3) 00:0d.0 IDE interface: nVidia Corporation MCP51 IDE (rev a1) 00:0e.0 IDE interface: nVidia Corporation MCP51 Serial ATA Controller (rev a1) 00:10.0 PCI bridge: nVidia Corporation MCP51 PCI Bridge (rev a2) 00:10.2 Multimedia audio controller: nVidia Corporation MCP51 AC97 Audio Controller (rev a2) 00:14.0 Bridge: nVidia Corporation MCP51 Ethernet Controller (rev a3) 00:18.0 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] HyperTransport Technology Configuration 00:18.1 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Address Map 00:18.2 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] DRAM Controller 00:18.3 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Miscellaneous Control [root@k5di ~]#
And the video controller is a nVidia Corporation C51G [GeForce 6100]
(rev a2) what ever that might mean. Also the entire computer is full of nvidia things and the only one causing trouble is the video.
Karl: First, whatever Soyo board that you have, it certainly isn't the model number that you cited above.
Second, unlike video cards, onboard chips _usually_ rely on system RAM rather than onboard RAM. If that is the case here, you have to reserve some memory space for the use of the chip by using a mem=XXXX command at boot time. (xxxx is the difference between the installed RAM and that required by the video card). I've never dealt with one of these onboard critters, so I'll leave it to you to determine the exact form of the mem command. In particular, be careful about expressing the amount of memory -- I'm not certain whether it's in kilobytes, megabyte or whatever.
-- cmg
Gads, I thought that was what the kernel does. Why must I tell it to save RAM? Part of the Nvidia software is a kernel package which might take care of this. At least I do not have any kernel stuff like that in my grub.conf.
--- Frank Cox theatre@sasktel.net wrote:
On Mon, 29 Oct 2007 15:24:35 +0800 Ed Greshko Ed.Greshko@greshko.com wrote:
OK... By that I take it there is no one
universally accepted definition of
those terms. Otherwise, there would be no need
for all the different
flavors of the "free" licenses. GPLv2, GPLv3,
FreeBSD, etc...
Each one of those licenses serves a particular purpose or need. (Whether each one is actually required is a different question.)
With regard to Free Software in particular, in the modern computer industry, the generally accepted definition of Free Software (notice the capitalization) is the definition that is provided and promoted by the Free Software Foundation. Which is what I sent you to see.
It's a kind of an "industry standard definition", for lack of a better term.
Many industrial trades, for example, use otherwise common terms that mean something "special" in terms of their specific industry. Free Software is a good example of an otherwise common term that has a special meaning in the computer industry.
-- MELVILLE THEATRE ~ Melville Sask ~ http://www.melvilletheatre.com
--
Users are invited to look at the following pages for more resources/to gain a better understanding of the meanings of Open Source:
Open Source page: http://www.opensource.org/docs/osd
Licenses by Category: http://www.opensource.org/licenses/category
Regards,
Antonio
__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
On Mon, 2007-10-29 at 10:59 +0000, Chris Jones wrote:
There are also other linux distros that don't take the same line as fedora. Ubuntu being the obvious example, that does make it much easier for users to use things like the nvidia driver.
Which means that those wanting the easy step should use those distros, not try to change Fedora into being just another of the same thing.
Ed Greshko wrote:
Frank Cox wrote:
On Sun, 28 Oct 2007 21:29:09 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Except that the people providing the binary do have a reason to care if it works.
Until they decide that it's time to sell you another card and discontinue their binary blob for the model that you already have.
You mean like when a developer of an opensource project decides he doesn't want or can't continue to support the project and it closes down for lack of others picking up the mantle?
Of course nvidia doesn't do what you are suggesting. They have their legacy channel as in:
Latest Legacy GPU version (1.0-71xx series): 71.86.01 Latest Legacy GPU version (1.0-96xx series): 96.43.01
As I pointed out yesterday Nvidia is a HUGE company and they make many things besides a VGA card. They are aware that the VGA card be it in a motherboard or a pci card will not work on Linux. So they put some people to work to make drivers for Linux and they ran into problems with the Linux world. The kernel kept changing was one. But they kept at it and we have have good drivers for Linux.
Now the Linux world is mad because Nvidia will not share the source code. Has anyone asked for the source code? There is likely a rule at Nvidia that no software written by them goes out with source code. This is the Microsoft idea.
So let us count our blessings. Nvidia is spending big bucks to make sure their hardware runs with Linux.
on 10/29/2007 8:13 AM, Karl Larsen wrote:
Ed Greshko wrote:
Frank Cox wrote:
On Sun, 28 Oct 2007 21:29:09 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Except that the people providing the binary do have a reason to care if it works.
Until they decide that it's time to sell you another card and discontinue their binary blob for the model that you already have.
You mean like when a developer of an opensource project decides he doesn't want or can't continue to support the project and it closes down for lack of others picking up the mantle?
Of course nvidia doesn't do what you are suggesting. They have their legacy channel as in:
Latest Legacy GPU version (1.0-71xx series): 71.86.01 Latest Legacy GPU version (1.0-96xx series): 96.43.01
As I pointed out yesterday Nvidia is a HUGE company and they make
many things besides a VGA card. They are aware that the VGA card be it in a motherboard or a pci card will not work on Linux. So they put some people to work to make drivers for Linux and they ran into problems with the Linux world. The kernel kept changing was one. But they kept at it and we have have good drivers for Linux.
Now the Linux world is mad because Nvidia will not share the source
code. Has anyone asked for the source code? There is likely a rule at Nvidia that no software written by them goes out with source code. This is the Microsoft idea.
The rule is named "We want to make money from our work and revealing how we get our video cards to do what they do (open the source) would allow ATI and others to copy our work and cut into any profit'. ;-)
Beg. Complain. Threaten. Whine. Do as you wish. It is not going to happen.
And it is not anyones fault. It is not anyone picking on Linux. The same code(s) is/are not open to Apple or Microsoft either. Nvidia provides a generic driver to them. And it is up to the user to upgrade to the better Nvidia drivers if they want to do so.
Another name for the rule? It is also called 'Good Business'.
So let us count our blessings. Nvidia is spending big bucks to make
sure their hardware runs with Linux.
on 10/29/2007 8:22 AM, Chris Jones wrote:
Which means that those wanting the easy step should use those distros, not try to change Fedora into being just another of the same thing.
Exactly my point - Theres a distro out there for everyone.
So tell me this Chris. Why do they hang around here and complain? Nothing else to do? ;-)
On 29/10/2007, Les Mikesell lesmikesell@gmail.com wrote:
Frank Cox wrote:
There are many ways the situation could be improved for the user.
The proper way to improve the situation for the user is for Nvidia to abide by the GPL and provide their drivers in the normal and correct manner.
They have claimed that they can't because parts are licensed from other parties.
A fairly lame excuse - they could still free up what they own and make a massive difference to the community, who would then no doubt jump in and replace the proprietory parts - see recent Sun/Java for example. Given that the Nouveau guys have done quite a lot already, if NVidia released what they could under a free license, I am sure that in no time we'd have functional drivers.
J.
Somebody in the thread at some point said:
The rule is named "We want to make money from our work and revealing how we get our video cards to do what they do (open the source) would allow ATI and others to copy our work and cut into any profit'. ;-)
Beg. Complain. Threaten. Whine. Do as you wish. It is not going to happen.
I don't really see that -- AMD / ATI supporting FOSS drivers now puts nVidia in an increasingly lonesome spot. I'll be avoiding nVidia like the plague until they also support FOSS drivers, even though in the past I would certainly have said nVidia would probably be a better choice when ATI / AMD was also closed.
The nice experiences I have with Intel laptops doing 3D with X with no meddling or crashes brings it close to my mind how important it is to have a fully capable FOSS driver.
Apparently the concern historically has been explained that they will lay themselves open to patent attacks by showing in the source what they are actually doing. But ATI / AMD don't seem to feel that is a problem... they probably have a sackful of bogus patents to whack nVidia over the head with if there was trouble, and the same in the other direction.
-Andy
Somebody in the thread at some point said:
On Mon, 2007-10-29 at 10:59 +0000, Chris Jones wrote:
There are also other linux distros that don't take the same line as fedora. Ubuntu being the obvious example, that does make it much easier for users to use things like the nvidia driver.
Which means that those wanting the easy step should use those distros, not try to change Fedora into being just another of the same thing.
Yes... there are advantages to having a distro where everything in there is redistributable no problems. (Other than you should probably offer people the source DVD at the same time as the binary one, to avoid the possibility of someone asking for it three years later.)
-Andy
David Boles wrote:
If you find a Linux distribution, a free one, that is different from what your attitude describes will you post the name here?
Most of my machines run Centos, where the interface doesn't change every week.
What is your latest disaster this time anyway? I missed that part of this thread.
My only recent ones have been the kernels that wouldn't boot on IBM MPT scsi controllers. I gave up on fedora on any machine with nvidia or firewire long ago.
Is that "IBM MPT scsi controllers" only? That does sound like a possible bug. Did you write a ticket on that?
There were 2 similar issues in recent months - I think one was a larger problem and was fixed in a few days. It may not have affected adaptec controllers. https://bugzilla.redhat.com/show_bug.cgi?id=251577 But how can things like that get pushed out with no testing? Most of my machines (obviously not running fedora...) are in remote locations where failing to boot is a serious problem.
Strange. Which version (release) of Fedora? Core 6? Fedora 7? I ask because I have an Nvidia graphics card and the firewire on my machine.
This was FC6 which had otherwise been stable for months.
I was using Nvidia graphics cards when I first tried Linux at RedHat v5.2. I think it was v5.2. It came with a "Linux for Dummies" book. I built my current machine around the time of Fedora 7. I, personally, do not have anything firewire but my son's camcorder works when attached. It did in Fedora 7 and still does in Fedora 8.
I use external firewire drives to periodically mirror some internal drives for an offsite backup copy. It worked in FC1 and about 25% of the time since. So fedora has just not been usable for anything important for years. And I am mildly amused at the claims of how much better open source drivers are...
Alan Cox wrote:
There's no piracy involved here as nvidia gives the drivers away and permits redistribution. It is the Linux side that makes their product
If I give away copies of someone elses movies and permit redistribution then thats still piracy.
How so, when the 'someone else' explicitly allows redistribution without modifications? Specifically:
"2.1.2 Linux/FreeBSD Exception. Notwithstanding the foregoing terms of Section 2.1.1, SOFTWARE designed exclusively for use on the Linux or FreeBSD operating systems, or other operating systems derived from the source code to these operating systems, may be copied and redistributed, provided that the binary files thereof are not modified in any way (except for unzipping of compressed files)."
The question is one of derived works and what the derived work boundary is.
Interfaces are interfaces.
Alan Cox wrote:
Then, even if you do make it, you'll have to establish why this isn't ordinary fair use of an interface required to interoperate with the kernel.
Because it's contrary to the license under which the Linux kernel is distributed. I don't think it can be made much clearer than that, frankly.
It's not at all clear why the kernel license should control someone else's work.
If the work is derived it isn't someone elses work, any more than if you add a scene to a movie and redistribute it. Thats the fundamental question - is the Nvidia driver a derived work. Some people believe no, some believe yes, nobody "knows" in the US legal sense because nobody has been to court to find out - and I suspect neither side considers it worth finding out.
I don't think you want to find out that a driver is derived from the OS it is written for, or you'll find all kinds of lawsuits crawling out of the closet about code included in linux that was originally written under some other OS. SCO's claims to that effect weren't really disproven since they were mostly dismissed on other counts.
On Sun October 28 2007, Ric Moore wrote:
You know me, Claude ...I bitched for months trying to figure out what was wrong, before finally getting Croquet to run. When the solution appeared, I was floored for wasting those months. While I'm grateful that nVidia supports linux and offer a driver package, I hope that nVidia will open their code, so this kinda stuff doesn't happen to anyone else. I seriously doubt that they would lose one bit of their sales. Ric
I would be curious to know the answer to that one. The implication might be that the Livna packager is removing something from the nVidia package before making the Fedora rpm...but, that may not be right. Since the guts of the nVidia driver is a binary, I don't understand what the packager could be removing -- I'm sure there's something I don't know in all this. Learning the answer would expand my knowledge.
Have you tried the packages from freshrpm's?
I run nVidia cards in 5 of my 6 Linux boxes, and I use the freshrpm driver in all. It's a done-once installation because the Dell dkms package takes over after that, and automatically builds a new kernel-module on the fly as I'm rebooting after each kernel upgrade. I have noted over time, that several people assert that the tar-balls from nVidia are superior to the rpm's that are available from Livna (there's very little discussion of the freshrpm variant on this list) - in my experience, that has not been the case, but, I don't use croquet, for example.
So far as Les' argument, I don't have a strong view. I can see both sides. For most of my friends who are not real interested in computing as a hobby, but who decide to try Linux, I generally put PCLinuxOS on their machines. PCLinuxOS does make one-click installation of nVidia drivers possible. They do much else that makes the experience much easier for the novice. (That happens to be their stated goal, a 'Windows-like' experience)
I prefer Fedora for many reasons. AND, I happen to respect their position on the free software question. I think that there's room for both positions in linux-land. I think there's a strong case to be made for a significant faction in the Linux community to stick their free-only guns -- it helps drive the whole open-source movement, and the general goals of that movement I do support, especially the development of open-source superior alternatives to all proprietary software. The devil is in the details...how to get there from where we are. There's room for different approaches as I see things right now.
I don't think you want to find out that a driver is derived from the OS it is written for, or you'll find all kinds of lawsuits crawling out of the closet about code included in linux that was originally written under some other OS.
Being written under some other OS is not the same as being derived from (any more than being compatible with is derived from). Derivative works are a much more complex idea than that which is one reason there is a lack of certainty on the issue.
Alan
On Mon, 29 Oct 2007 08:12:36 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Alan Cox wrote:
There's no piracy involved here as nvidia gives the drivers away and permits redistribution. It is the Linux side that makes their product
If I give away copies of someone elses movies and permit redistribution then thats still piracy.
How so, when the 'someone else' explicitly allows redistribution without modifications? Specifically:
The "someone else" is the Linux community. Nvidia can't give away my rights they don't own any more than I can give away Nvidia's rights.
Alan
Now the Linux world is mad because Nvidia will not share the source
code. Has anyone asked for the source code? There is likely a rule at
Actually people have repeatedly asked for chip documentation. The code (as with ATI) is almost certainly going to contain third party components they can't publish in source form - probably nowdays owned by Microsoft rather than SGI.
As it is Nouveau is having to reverse engineer all the hardware details to support the chips with free software. Thats a hard and tedious process.
For the ATI driver we now have documentation, not ATI source code, and that will lead to 3D drivers. Intel write their own drivers, VIA sometimes throw stuff over the wall and run away which is how we got VIA 3D for some of their chipsets.
Nvidia that no software written by them goes out with source code. This is the Microsoft idea.
About 10 seconds of investigation before typing would have shown this untrue. Reading previous emails likewise where I pointed out that Nvidia wrote and supplied the open source 2D Nvidia driver we ship as well...
Alan
Alan Cox wrote:
If I give away copies of someone elses movies and permit redistribution then thats still piracy. The question is one of derived works and what the derived work boundary is.
I think Alan Cox - whom I greatly admire for his many contributions to Linux - is being disingenuous in concentrating on the legal aspects of nVidia drivers.
Some obscure legal issue may arise, but the fact of the matter is nVidia's closed drivers would not be acceptable to Fedora with or without legalities.
In my (limited) experience closed drivers create serious problems, as you never know if they will work with newer kernels or other software, so you cannot rely on them too much.
On Mon, 29 Oct 2007, Alan Cox wrote:
Nvidia that no software written by them goes out with source code. This is the Microsoft idea.
About 10 seconds of investigation before typing would have shown this untrue. Reading previous emails likewise where I pointed out that Nvidia wrote and supplied the open source 2D Nvidia driver we ship as well...
"adventures in fedoraland: episode 437", in which uber-geek alan cox learns that being "output only" doesn't just apply to hardware devices.
rday
Alan Cox wrote:
There's no piracy involved here as nvidia gives the drivers away and permits redistribution. It is the Linux side that makes their product
If I give away copies of someone elses movies and permit redistribution then thats still piracy.
How so, when the 'someone else' explicitly allows redistribution without modifications? Specifically:
The "someone else" is the Linux community. Nvidia can't give away my rights they don't own any more than I can give away Nvidia's rights.
As soon as you can prove that your code is included in nvidia's module, you'll have the right to say that. Or at least more code than is necessary for fair use of the interface provided by the kernel.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Everyone,
(1) This topic has been hashed, re-hashed and mulled over for many years now. Please do everyone a favor and STOP!
(2) The release or non-release of drivers, source, or other documentation is up to any manufacturer of product. If they choose NOT to release, than that is a business decision they have made. The only valid complaint anyone can make with such a decision is with our MONEY. If Nvidia stops getting money for their product, it poses a RED flag to the company the market has changed. In the end, MONEY speaks louder than words to these companies.
Thanks, James
On Sun, 2007-10-28 at 22:11 -0500, Les Mikesell wrote:
Frank Cox wrote:
Except that the people providing the binary do have a reason to care if it works.
Until they decide that it's time to sell you another card and discontinue their binary blob for the model that you already have.
In a reasonable OS, a binary blob that worked today would work again tomorrow.
I have gone from Fedora 5 to 6 to 7 with the same nvidia blob. Occasionally the fix for bugs or security will no doubt affect memory allocation, driver argument processing, or the manner of input from peripheral devices, and this in turn may require some change in the binary driver to be consistant. However, this is not a fault of either OS or binary driver, but rather a failing of human nature, and one that I, as a programmer can tell you takes decades of experience to even minimize, forget about eliminating it, because the fault you don't know or recognize you cannot code to prevent.
Regards, Les H
On Sun, 2007-10-28 at 20:14 -0700, Antonio Olivares wrote:
--- Les Mikesell lesmikesell@gmail.com wrote:
Antonio Olivares wrote:
Fedora will never take responsibility for closed source software/drivers
not
working correctly. You are at your own risk.
Wait - are you saying that they _do_ take responsibility for the open source they distribute?
I am not implying that, but since you conclude that on your own, the upstream developers and the fedora developers work hand in hand to fix code and release fixes for them. But for nvidia code and drivers they are not to take any blame for the success/failure of the binary drivers. That's why there is the "use at your risk" part in the statement.
Who should I call the next time my firewire drives aren't recognized or an update kernel won't boot?
You can do a number of things that might/might not get answered (1) File Bug Reports (2) File RFE's (3) complain, complain, complain, and hope that someone listens and tries to help (4) wait patiently till a fix makes its way via updates
How soon will someone be over to fix it?
Depends on the issues and the severity of the bugs. Developers work hard and try to release updates to fix problems.
-- Les Mikesell lesmikesell@gmail.com
--
Regards,
Antonio
You left out getting the sources for the OS and submitting a fix for the interface, or comparing the version with which the driver worked and the version where the driver doesn't work and telling the development team to help isolate the problem.
Regards, Les H
Frank Cox wrote:
On Sun, 28 Oct 2007 22:11:29 -0500 Les Mikesell lesmikesell@gmail.com wrote:
In a reasonable OS, a binary blob that worked today would work again tomorrow.
I think the word you're looking for is moribund.
Frank Cox is one of many out of touch with reality. The Reality is that 95% of your friends use Windows and the send you stuff they really like. If you stick with Frank you will never see or hear those things. The pure Fedora viewer and music player are bad. I have VLC with the codex library which is illegal and will get me 20 years in jail. At 72 years old that would about do it. But VLC pays DVD movies and every kind of file known to Windows. I would have Windows on this computer if it was not for VLC.
I would not be on this F7 if it were not for the Nvidia drivers they make available. And you Frank hate them. Because they do not provide you with their source code.
Another thing you hate is Thunderbird because it does not ship with it's source code. But if you took a pole of Fedora users today you will find a large percentage are using T-Bird.
Be reasonable. There are things in the world better than what Fedora can produce and these things that are free should be embraced not spit upon.
Timothy Murphy wrote:
If I give away copies of someone elses movies and permit redistribution then thats still piracy. The question is one of derived works and what the derived work boundary is.
I think Alan Cox - whom I greatly admire for his many contributions to Linux - is being disingenuous in concentrating on the legal aspects of nVidia drivers.
Some obscure legal issue may arise, but the fact of the matter is nVidia's closed drivers would not be acceptable to Fedora with or without legalities.
In my (limited) experience closed drivers create serious problems, as you never know if they will work with newer kernels or other software, so you cannot rely on them too much.
In my experience, open source drivers have as many or more problems than the closed source versions. _If_ you are Alan Cox or some number of hackers with equivalently specialized skills that you could probably count on one hand, having the source code available might be of some value when the supplied binary doesn't work. The rest of us report the bug and wait, and again in my experience over the last couple of decades, the closed source providers are at least equally responsive in this scenario.
On Mon, 29 Oct 2007, Karl Larsen wrote:
Another thing you hate is Thunderbird because it does not ship with it's source code.
gawd almighty, karl, do you do *any* research before you post such drivel? it took me a single (fedora) command to download the source rpm for thunderbird:
$ yumdownloader --source thunderbird
but wait ... there's more!
$ rpm -qi thunderbird Name : thunderbird Relocations: (not relocatable) Version : 2.0.0.6 Vendor: Fedora Project Release : 6.fc8 Build Date: Wed 26 Sep 2007 11:59:11 AM EDT Install Date: Mon 29 Oct 2007 10:51:03 AM EDT Build Host: xenbuilder2.fedora.redhat.com Group : Applications/Internet Source RPM: thunderbird-2.0.0.6-6.fc8.src.rpm Size : 46130946 License: MPLv1.1 or GPLv2+ or LGPLv2+ Signature : DSA/SHA1, Thu 25 Oct 2007 10:04:57 AM EDT, Key ID b44269d04f2a6fd2 Packager : Fedora Project URL : http://www.mozilla.org/projects/thunderbird/ Summary : Mozilla Thunderbird mail/newsgroup client Description : Mozilla Thunderbird is a standalone mail and newsgroup client.
and look! over there!
"Like its sibling Firefox, the beauty of the open source Thunderbird ^^^^^^^^^^^ email client is its extensibility. Sure we love our web-based email like Gmail, but Thunderbird is the ultimate open source desktop email ^^^^^^^^^^^ app."
in response to the above, i'm guessing karl will call me "stupid."
rday
Somebody in the thread at some point said:
I would not be on this F7 if it were not for the Nvidia drivers they make available. And you Frank hate them. Because they do not provide you with their source code.
Another thing you hate is Thunderbird because it does not ship with it's source code. But if you took a pole of Fedora users today you will find a large percentage are using T-Bird.
Nonsense. You can get it from Mozilla here
ftp://ftp.mozilla.org/pub/mozilla.org/thunderbird/releases/2.0.0.6/source/
or from Fedora itself here for example
http://koji.fedoraproject.org/packages/thunderbird/2.0.0.5/2.fc7/src/thunder...
-Andy
On Mon, 2007-10-29 at 08:44 -0600, Karl Larsen wrote:
Frank Cox wrote:
On Sun, 28 Oct 2007 22:11:29 -0500 Les Mikesell lesmikesell@gmail.com wrote:
In a reasonable OS, a binary blob that worked today would work again tomorrow.
I think the word you're looking for is moribund.
Frank Cox is one of many out of touch with reality. The Reality is that 95% of your friends use Windows and the send you stuff they really like. If you stick with Frank you will never see or hear those things. The pure Fedora viewer and music player are bad. I have VLC with the codex library which is illegal and will get me 20 years in jail. At 72 years old that would about do it. But VLC pays DVD movies and every kind of file known to Windows. I would have Windows on this computer if it was not for VLC.
I think the point Frank and others are making (though it's hard to tell, based on the one sentence you've quoted) is that Fedora *can't* include open-source but patented codecs without risking a lawsuit because it is based in the US.
I would not be on this F7 if it were not for the Nvidia drivers
they make available. And you Frank hate them. Because they do not provide you with their source code.
A totally separate issue from the one mentioned above. With the nVidia drivers, the problem is the lack of source code, which means that including it would violate Fedora's guidelines. Whatever your position is on nVidia's drivers, the fact remains that Fedora isn't going to change its guidelines.
Another thing you hate is Thunderbird because it does not ship with
it's source code. But if you took a pole of Fedora users today you will find a large percentage are using T-Bird.
Yes, because, cunningly enough, the Thunderbird developers have decided to release the Thunderbird code to the world under an open source license, much like the rest of the software in Fedora.
Be reasonable. There are things in the world better than what Fedora
can produce and these things that are free should be embraced not spit upon.
I don't think phlegm is the issue. I think the real issue is that some would like Fedora to change its spots, and (fortunately or unfortunately, depending on your position) it's not going to happen.
Jonathan
Frank Cox wrote:
If I depend on say "Oracle" and since Oracle is proprietary I should not run it on Linux?
You are taking a general statement and applying it to a very specific circumstance.
In the particular case cited, Linux may indeed be the answer you seek. However, in general terms, the solutions that folks like Les are looking for can be most easily found in places other than Linux. Microsoft and Apple are the two largest examples of such places.
Each of those introduce their own issues: OS X only runs on a limited set of hardware and we all know about the problems with Microsoft. However, in my opinion, the GPL, by preventing competitive combinations of software components, has made more money for Microsoft than anything Steve Balmer has ever said or done.
on 10/29/2007 9:03 AM, Les Mikesell wrote:
David Boles wrote:
If you find a Linux distribution, a free one, that is different from what your attitude describes will you post the name here?
Most of my machines run Centos, where the interface doesn't change every week.
What is your latest disaster this time anyway? I missed that part of this thread.
My only recent ones have been the kernels that wouldn't boot on IBM MPT scsi controllers. I gave up on fedora on any machine with nvidia or firewire long ago.
Is that "IBM MPT scsi controllers" only? That does sound like a possible bug. Did you write a ticket on that?
There were 2 similar issues in recent months - I think one was a larger problem and was fixed in a few days. It may not have affected adaptec controllers. https://bugzilla.redhat.com/show_bug.cgi?id=251577 But how can things like that get pushed out with no testing? Most of my machines (obviously not running fedora...) are in remote locations where failing to boot is a serious problem.
Strange. Which version (release) of Fedora? Core 6? Fedora 7? I ask because I have an Nvidia graphics card and the firewire on my machine.
This was FC6 which had otherwise been stable for months.
I was using Nvidia graphics cards when I first tried Linux at RedHat v5.2. I think it was v5.2. It came with a "Linux for Dummies" book. I built my current machine around the time of Fedora 7. I, personally, do not have anything firewire but my son's camcorder works when attached. It did in Fedora 7 and still does in Fedora 8.
I use external firewire drives to periodically mirror some internal drives for an offsite backup copy. It worked in FC1 and about 25% of the time since. So fedora has just not been usable for anything important for years. And I am mildly amused at the claims of how much better open source drivers are...
Just curious. The firewire drive problem. Are you staing that what you say, that they don't work well, is *all* Fedora and that they do work well with other distributions? Or that they don't work with the kernels, same number, of all of them?
Honestly if these drives were as important to me as you say that they are for you and that they worked with a different distro I would have switched a long, long time ago.
on 10/29/2007 11:09 AM, Les Mikesell wrote:
Frank Cox wrote:
If I depend on say "Oracle" and since Oracle is proprietary I should not run it on Linux?
You are taking a general statement and applying it to a very specific circumstance.
In the particular case cited, Linux may indeed be the answer you seek. However, in general terms, the solutions that folks like Les are looking for can be most easily found in places other than Linux. Microsoft and Apple are the two largest examples of such places.
Each of those introduce their own issues: OS X only runs on a limited set of hardware and we all know about the problems with Microsoft. However, in my opinion, the GPL, by preventing competitive combinations of software components, has made more money for Microsoft than anything Steve Balmer has ever said or done.
"the problems with Microsoft"? What would those be Les? List of problems please.
Today Les Mikesell did spake thusly:
Rahul Sundaram wrote:
Les Mikesell wrote:
There are many ways the situation could be improved for the user. The best I would hope for would be the inclusion of a configuration for a yum repository containing the drivers currently maintained in the livna repo. If that's too much to ask, then a link to the repo from the fedora project site would be better than nothing. Actually including the driver or maintaining a stable interface would be too much to expect here.
Again, Fedora will only include Free and open source software. Supporting non-free software is any manner is a explicit non-objective.
We all know that the real objective of fedora is to get a lot of users to do the beta testing on what will be the next RHEL after which they quickly stop supporting those tested versions and move on. Maybe they get enough testers without having to be bothered making it convenient for nvidia, vmware, etc. users.
If that were true then my FC6 media server running an old NVIDIA card I had kicking around probably wouldn't work, would it?
Bloody drama queens ;)
FWIW, I've always used Nvidia (cept my laptop, which is Intel) and I've never had a problem with any of it, cept for dev versions of compiz/XGL not quite working right when I was playing a year or so ago.
If you want to run linux, you do some investigation online. Google is your friend. It's pretty easy to set up the nvidia drivers, the nv drivers are included out of the box so it'll all work anyway, the only thing that you won't get is 3d acceleration.
Frank Cox wrote:
Do you not recognize that there are other existing licenses on technology that users need?
There are all kinds of licenses. If your requirements include "proprietary, closed-source and immutable software", then Linux is not for you.
Linus was widely quoted as saying the opposite in the early days of Linux. I doubt if it would have had any acceptance otherwise.
I'm sorry if this is the first time you have heard this.
There's obviously a misunderstanding somewhere. One piece of software does not get to dictate what others you can run regardless of the unrealistic fanaticism involved. But it does make opensolaris sound more appealing.
Today Anders Karlsson did spake thusly:
Thus, Les Mikesell at Mon Oct 29 03:11:29 2007 inscribed:
Frank Cox wrote:
Except that the people providing the binary do have a reason to care if it works.
Until they decide that it's time to sell you another card and discontinue their binary blob for the model that you already have.
In a reasonable OS, a binary blob that worked today would work again tomorrow.
Les,
If you have such a beef about this particularly part about the argument, I suggest you take it onto the Linux Kernel mailing list. I would suggest you search the archives for it first though, as it has been debated extensively already and all the arguments you can come up with most likely have been replied to. It'll save traffic on this list.
Just an observation...
The point Les appears to be missing is that the binary blob doesn't stay the same because the cards it supports don't stay the same. And as NVidia upgrade their hardware, they'll slowly stop supporting the older hardware.
A few years ago a nice man called Konst made something called centericq. It supported lots of nice IM protocols and was pretty much perfect. He released the source code under a GPL license and then decided after a while to stop maintaining it.
Because of the GPL, some other people decided to pick it up, rename it centerim and change it, so when the evil microsoft changed their IM protocols again it would still work.
This is what we call "free" software. IF NVidia stop supporting anything below a series 5 GPU, no-one can step in and offer support as there's no source code and no license allowing them to do this.
So. Whilst the original binary blob will be fine and dandy for a while, there may be security issues discovered around it, there may be an overhaul of how graphics should work that mean you could get twice the power out of the card. There might be all kinds of things that change, not least of all the kernel interface to take into account some of these things.
Progress isn't a bad thing. Ranting at Fedora people because someone else's hardware isn't easy to make work on their OS is a bit silly though.
Would you moan at redhat if you couldn't get Oracle to work on it, yet Oracle claimed linux compatibility?
David Boles wrote:
Just curious. The firewire drive problem. Are you staing that what you say, that they don't work well, is *all* Fedora and that they do work well with other distributions? Or that they don't work with the kernels, same number, of all of them?
In fedora, whether they work or not changes as fast as the weather in Chicago. Other distributions don't push changes out to users just to see what happens. In more sensible times, there was an explicit 'unstable' branch for kernel development so it was clear what had a chance of working in a distribution. Now all the wild and crazy changes are thrown in the same pot and it is left up to each distribution to guess what is ready for prime time.
Honestly if these drives were as important to me as you say that they are for you and that they worked with a different distro I would have switched a long, long time ago.
I did - and the Centosplus kernels have not given me any unpleasant surprises.
Scott van Looy wrote:
Progress isn't a bad thing. Ranting at Fedora people because someone else's hardware isn't easy to make work on their OS is a bit silly though.
But, but, but... What's the point of an OS if it doesn't make it easy to use hardware????
Would you moan at redhat if you couldn't get Oracle to work on it, yet Oracle claimed linux compatibility?
Of course. There's not much excuse for screwing up interfaces that an OS is supposed to provide.
On Mon, Oct 29, 2007 at 09:56:42 -0500, Les Mikesell lesmikesell@gmail.com wrote:
In my experience, open source drivers have as many or more problems than the closed source versions. _If_ you are Alan Cox or some number of hackers with equivalently specialized skills that you could probably count on one hand, having the source code available might be of some value when the supplied binary doesn't work. The rest of us report the bug and wait, and again in my experience over the last couple of decades, the closed source providers are at least equally responsive in this scenario.
The other category of people helped are those with enough money to hire a driver expert to fix the driver for them. That doesn't cover (almost all) one off users. But large enough organizations could potentially afford this.
Today Les Mikesell did spake thusly:
Scott van Looy wrote:
Progress isn't a bad thing. Ranting at Fedora people because someone else's hardware isn't easy to make work on their OS is a bit silly though.
But, but, but... What's the point of an OS if it doesn't make it easy to use hardware????
To operate the CPU(s) - hardware support is optional and often in obscure places absent or wonky at best. Device drivers are the single thing that causes more problems with _all_ OSen than anything else. If my webcam didn't work under vista I'd blame the webcam manufacturer for not having created drivers that work, not Microsoft for not having anticipated a webcam of a certain type.
Would you moan at redhat if you couldn't get Oracle to work on it, yet Oracle claimed linux compatibility?
Of course. There's not much excuse for screwing up interfaces that an OS is supposed to provide.
And the fact that everyone at Redhat would tell you to talk to Oracle for writing broken software wouldn't dissuade you?
On Mon, Oct 29, 2007 15:29:46 PM +0800, Ed Greshko (Ed.Greshko@greshko.com) wrote:
Frank Cox wrote:
If you depend on proprietary software, then Linux is not for you. Microsoft and/or Apple products are a much better fit for your needs.
Frank,
this is the second time (as far as I can tell now) that you insist on this point in this thread, but Ed is right:
Never did care that much for blanket statements. If I depend on say "Oracle" and since Oracle is proprietary I should not run it on Linux?
I don't know much about big databases, but CAD/CAE is another field where the best software right now is proprietary _and_ runs much better on Linux for lots of reasons. Including software that needs high-end graphics, not just server/CLI apps.
Ciao, Marco
Scott van Looy wrote:
Progress isn't a bad thing. Ranting at Fedora people because someone else's hardware isn't easy to make work on their OS is a bit silly though.
But, but, but... What's the point of an OS if it doesn't make it easy to use hardware????
To operate the CPU(s) - hardware support is optional and often in obscure places absent or wonky at best.
If your i/o devices don't work, whether your CPU works or not becomes a Heisnberg question. And the answer won't really matter.
Would you moan at redhat if you couldn't get Oracle to work on it, yet Oracle claimed linux compatibility?
Of course. There's not much excuse for screwing up interfaces that an OS is supposed to provide.
And the fact that everyone at Redhat would tell you to talk to Oracle for writing broken software wouldn't dissuade you?
I might peek at some strace output to see whether the system call arguments or results were at fault.
Bruno Wolff III wrote:
In my experience, open source drivers have as many or more problems than the closed source versions. _If_ you are Alan Cox or some number of hackers with equivalently specialized skills that you could probably count on one hand, having the source code available might be of some value when the supplied binary doesn't work. The rest of us report the bug and wait, and again in my experience over the last couple of decades, the closed source providers are at least equally responsive in this scenario.
The other category of people helped are those with enough money to hire a driver expert to fix the driver for them. That doesn't cover (almost all) one off users. But large enough organizations could potentially afford this.
The issue is theoretical at best. In the unlikely event that access to a video card breaks due to undiscovered bugs in the original _and_ vendor refusal to fix it, I'd expect it to be cheaper to either replace Linux or the card than to hire an expert to temporarily revive the now-dead combination.
Les Mikesell wrote:
Scott van Looy wrote:
Progress isn't a bad thing. Ranting at Fedora people because someone else's hardware isn't easy to make work on their OS is a bit silly though.
But, but, but... What's the point of an OS if it doesn't make it easy to use hardware????
Lots of hardware just works. It's vendors that cannot or will not share the specs of their cards that have trouble. The best solution would be to complain to the vendor to open up their specs.
Would you moan at redhat if you couldn't get Oracle to work on it, yet Oracle claimed linux compatibility?
Of course. There's not much excuse for screwing up interfaces that an OS is supposed to provide.
Fedora's goal is to stay as close to upstream as possible and provide very current versions. For the kernel, this means that the interface does change sometimes. It is a simple fact. If you don't like it, please take your case to LKML. It is ridiculous, pointless, counterproductive, and IMO off-topic to bring it up here -- repeatedly.
On Mon, 2007-10-29 at 03:21 -0400, Ric Moore wrote:
On Sun, 2007-10-28 at 20:05 -0700, Craig White wrote:
On Sun, 2007-10-28 at 22:37 -0400, David Boles wrote:
What is your latest disaster this time anyway? I missed that part of this thread.
Les' mission is to save us from the evils of GPL license.
Not so, no where did Les say that, Craig. I know Les ...he has a bunch of coding chops under his belt and he's a helluva decent human being. The point is that it needn't be anathema if the only way to get a device working correctly is to use a driver from the manufacturer. Sure, pang on them until the cows come home to open up their code. I believe in that. One day everyone will see * * The Light * *. Guys like Alan Cox could tell them a thing or three or ten about proper interfacing to the kernel.
Meanwhile, getting stuff to work is a consideration for Joe Lunch de Bucket, Les, me and a bunch of others. And, we deal with what we got as we can. :) Ric
---- evidently you don't know Les all that well or simply haven't tracked his feelings about GPL
Craig
On Sun, 2007-10-28 at 23:43 -0600, Frank Cox wrote:
On Mon, 29 Oct 2007 00:28:48 -0500 Les Mikesell lesmikesell@gmail.com wrote:
Do you not recognize that there are other existing licenses on technology that users need?
There are all kinds of licenses. If your requirements include "proprietary, closed-source and immutable software", then Linux is not for you.
I'm sorry if this is the first time you have heard this.
With all due respect, sir, you don't know jack about my requirements, limitations, and objectives, or what environment I work in, or what tools are available in my field, or what I may or may not be in a position to do about it.
But feel free to pass judgment anyway. After all, it's a simple world out there, black and white are all you need to know. No room here for nuance or complexity. If I'm not 100% with you, then I must be with the enemy.
Pass the kool-aid...
Somebody in the thread at some point said:
Bruno Wolff III wrote:
In my experience, open source drivers have as many or more problems than the closed source versions. _If_ you are Alan Cox or some number of hackers with equivalently specialized skills that you could probably count on one hand, having the source code available might be of some value when the supplied binary doesn't work. The rest of us report the bug and wait, and again in my experience over the last couple of decades, the closed source providers are at least equally responsive in this scenario.
The other category of people helped are those with enough money to hire a driver expert to fix the driver for them. That doesn't cover (almost all) one off users. But large enough organizations could potentially afford this.
The issue is theoretical at best. In the unlikely event that access to a video card breaks due to undiscovered bugs in the original _and_ vendor refusal to fix it, I'd expect it to be cheaper to either replace Linux or the card than to hire an expert to temporarily revive the now-dead combination.
Well whatever your other complaints, I really don't think you take into account the developer suffering that happens from the unsupported reverse engineering aspect that is often part of the drivers. Sometimes folk don't even get thrown a datasheet from the manufacturer. You just stare at hundred and even thousands of 32 bit registers with twinkling mystical unknown purposes and try to figure out how to make it do something useful.
It's like the proverbial dog playing the piano, it's not that it didn't play it perfectly (although it often does in my experience) but that it can play it at all that is remarkable. I looked through lshal the other day and it was listing dozens and dozens of devices in this laptop that had FOSS drivers.
More than that though I myself have taken advantage of a kernel driver blowing a panic to look through the source and fix the problem, and send a patch describing and fixing to problem, which was accepted. I was really excited by being able to do that and still think that system must most times knock crap out of doing it in-house with proprietary code.
As more manufacturers support external efforts with data quality and completeness will definitely generally rise.
-Andy
Robert P. J. Day wrote:
On Mon, 29 Oct 2007, Karl Larsen wrote:
Another thing you hate is Thunderbird because it does not ship with it's source code.
gawd almighty, karl, do you do *any* research before you post such drivel? it took me a single (fedora) command to download the source rpm for thunderbird:
$ yumdownloader --source thunderbird
but wait ... there's more!
$ rpm -qi thunderbird Name : thunderbird Relocations: (not relocatable) Version : 2.0.0.6 Vendor: Fedora Project Release : 6.fc8 Build Date: Wed 26 Sep 2007 11:59:11 AM EDT Install Date: Mon 29 Oct 2007 10:51:03 AM EDT Build Host: xenbuilder2.fedora.redhat.com Group : Applications/Internet Source RPM: thunderbird-2.0.0.6-6.fc8.src.rpm Size : 46130946 License: MPLv1.1 or GPLv2+ or LGPLv2+ Signature : DSA/SHA1, Thu 25 Oct 2007 10:04:57 AM EDT, Key ID b44269d04f2a6fd2 Packager : Fedora Project URL : http://www.mozilla.org/projects/thunderbird/ Summary : Mozilla Thunderbird mail/newsgroup client Description : Mozilla Thunderbird is a standalone mail and newsgroup client.
and look! over there!
"Like its sibling Firefox, the beauty of the open source Thunderbird ^^^^^^^^^^^ email client is its extensibility. Sure we love our web-based email like Gmail, but Thunderbird is the ultimate open source desktop email ^^^^^^^^^^^ app."
in response to the above, i'm guessing karl will call me "stupid."
rday
Shame on you rday, your using a non-fedora email system. Say what now?
On Mon, 2007-10-29 at 11:45 -0600, Karl Larsen wrote:
Robert P. J. Day wrote:
On Mon, 29 Oct 2007, Karl Larsen wrote:
Another thing you hate is Thunderbird because it does not ship with it's source code.
gawd almighty, karl, do you do *any* research before you post such drivel? it took me a single (fedora) command to download the source rpm for thunderbird:
$ yumdownloader --source thunderbird
but wait ... there's more!
$ rpm -qi thunderbird Name : thunderbird Relocations: (not relocatable) Version : 2.0.0.6 Vendor: Fedora Project Release : 6.fc8 Build Date: Wed 26 Sep 2007 11:59:11 AM EDT Install Date: Mon 29 Oct 2007 10:51:03 AM EDT Build Host: xenbuilder2.fedora.redhat.com Group : Applications/Internet Source RPM: thunderbird-2.0.0.6-6.fc8.src.rpm Size : 46130946 License: MPLv1.1 or GPLv2+ or LGPLv2+ Signature : DSA/SHA1, Thu 25 Oct 2007 10:04:57 AM EDT, Key ID b44269d04f2a6fd2 Packager : Fedora Project URL : http://www.mozilla.org/projects/thunderbird/ Summary : Mozilla Thunderbird mail/newsgroup client Description : Mozilla Thunderbird is a standalone mail and newsgroup client.
and look! over there!
"Like its sibling Firefox, the beauty of the open source Thunderbird ^^^^^^^^^^^ email client is its extensibility. Sure we love our web-based email like Gmail, but Thunderbird is the ultimate open source desktop email ^^^^^^^^^^^ app."
in response to the above, i'm guessing karl will call me "stupid."
rday
Shame on you rday, your using a non-fedora email system. Say what now?
---- thunderbird is part of fedora base packages
Craig
On Mon, 29 Oct 2007 13:23:06 -0400 Matthew Saltzman mjs@CLEMSON.EDU wrote:
With all due respect, sir, you don't know jack about my requirements, limitations, and objectives, or what environment I work in, or what tools are available in my field, or what I may or may not be in a position to do about it.
You could always try open hostility. It's highly recommended as a great way to get others to assist you with your problems.
On Mon, 29 Oct 2007 13:00:36 +0000 Andy Green andy@warmcat.com wrote:
Apparently the concern historically has been explained that they will lay themselves open to patent attacks by showing in the source what they are actually doing.
That sounds like a very odd explanation, if indeed that is the way that they described it. "I don't want to tell you what I do for a living because I would have to admit that I'm a burglar."
It seems to me that statement like that could provide "probable cause" (or whatever the correct term is) for a patent holder to get a court order to have those guys checked out for patent violations.
Craig White wrote:
Meanwhile, getting stuff to work is a consideration for Joe Lunch de Bucket, Les, me and a bunch of others. And, we deal with what we got as we can. :) Ric
evidently you don't know Les all that well or simply haven't tracked his feelings about GPL
But it is all about getting things to work together - my feelings about the GPL are just a result of it's restrictions that prevent working and useful things from being distributed.
Andy Green wrote:
The issue is theoretical at best. In the unlikely event that access to a video card breaks due to undiscovered bugs in the original _and_ vendor refusal to fix it, I'd expect it to be cheaper to either replace Linux or the card than to hire an expert to temporarily revive the now-dead combination.
Well whatever your other complaints, I really don't think you take into account the developer suffering that happens from the unsupported reverse engineering aspect that is often part of the drivers.
Not only do I not take it into account, I can't understand why anyone thinks this is desirable compared to using drivers written and maintained by the engineers that build the hardware and have the test equipment to diagnose it.
More than that though I myself have taken advantage of a kernel driver blowing a panic to look through the source and fix the problem, and send a patch describing and fixing to problem, which was accepted.
Again, this doesn't sound like a desirable scenario compared to using something that already works.
Somebody in the thread at some point said:
Andy Green wrote:
The issue is theoretical at best. In the unlikely event that access to a video card breaks due to undiscovered bugs in the original _and_ vendor refusal to fix it, I'd expect it to be cheaper to either replace Linux or the card than to hire an expert to temporarily revive the now-dead combination.
Well whatever your other complaints, I really don't think you take into account the developer suffering that happens from the unsupported reverse engineering aspect that is often part of the drivers.
Not only do I not take it into account, I can't understand why anyone thinks this is desirable compared to using drivers written and maintained by the engineers that build the hardware and have the test equipment to diagnose it.
It's desirable when the device you want a driver for otherwise drags in Windows as a "dependency". That puts you into a situation where your choice of video card or whatever is making the decisions about security policy for you, and all the other areas that the choice of OS touches on.
More than that though I myself have taken advantage of a kernel driver blowing a panic to look through the source and fix the problem, and send a patch describing and fixing to problem, which was accepted.
Again, this doesn't sound like a desirable scenario compared to using something that already works.
You never had a closed source driver with a bug in? There's nothing for you to do but make a bug report and wait.
-Andy
David Boles wrote:
If I depend on say "Oracle" and since Oracle is proprietary I should not run it on Linux?
You are taking a general statement and applying it to a very specific circumstance.
In the particular case cited, Linux may indeed be the answer you seek. However, in general terms, the solutions that folks like Les are looking for can be most easily found in places other than Linux. Microsoft and Apple are the two largest examples of such places.
Each of those introduce their own issues: OS X only runs on a limited set of hardware and we all know about the problems with Microsoft. However, in my opinion, the GPL, by preventing competitive combinations of software components, has made more money for Microsoft than anything Steve Balmer has ever said or done.
"the problems with Microsoft"? What would those be Les? List of problems please.
Life is too short to complete that list... But start with the thousands of vulnerabilities that produced the estimated 25% of all current PCs being controlled in botnets that are used to send spam and DDOS attacks. Or the massive waste of time and money spent to protect the rest with 3rd party add-ons and frequent necessary grooming. Then there are the illegal anti-competitive marketing practices, the funding of SCO, the patent threats against open source. Binary device drivers are not part of this list, though. Since at least Win2ksp2 they basically just work.
Andy Green wrote:
You never had a closed source driver with a bug in? There's nothing for you to do but make a bug report and wait.
And that's what I would do with open source as well. The difference is that closed source OS's rarely change their driver interfaces, so it would be extremely unusual for something that already works and I have put into production to suddenly fail due to an update. In fedora, this has been a regular occurrence.
Les Mikesell wrote:
Andy Green wrote:
The issue is theoretical at best. In the unlikely event that access to a video card breaks due to undiscovered bugs in the original _and_ vendor refusal to fix it, I'd expect it to be cheaper to either replace Linux or the card than to hire an expert to temporarily revive the now-dead combination.
Well whatever your other complaints, I really don't think you take into account the developer suffering that happens from the unsupported reverse engineering aspect that is often part of the drivers.
Not only do I not take it into account, I can't understand why anyone thinks this is desirable compared to using drivers written and maintained by the engineers that build the hardware and have the test equipment to diagnose it.
More than that though I myself have taken advantage of a kernel driver blowing a panic to look through the source and fix the problem, and send a patch describing and fixing to problem, which was accepted.
Again, this doesn't sound like a desirable scenario compared to using something that already works.
I am on the Nvidia web page and here is there software license. Note the differant license for Linux:
License For Customer Use of NVIDIA Software
IMPORTANT NOTICE -- READ CAREFULLY: This License For Customer Use of NVIDIA Software ("LICENSE") is the agreement which governs use of the software of NVIDIA Corporation and its subsidiaries (“NVIDIA”) downloadable herefrom, including computer software and associated printed materials ("SOFTWARE"). By downloading, installing, copying, or otherwise using the SOFTWARE, you agree to be bound by the terms of this LICENSE. If you do not agree to the terms of this LICENSE, do not download the SOFTWARE.
RECITALS
Use of NVIDIA's products requires three elements: the SOFTWARE, the hardware on a graphics controller board, and a personal computer. The SOFTWARE is protected by copyright laws and international copyright treaties, as well as other intellectual property laws and treaties. The SOFTWARE is not sold, and instead is only licensed for use, strictly in accordance with this document. The hardware is protected by various patents, and is sold, but this LICENSE does not cover that sale, since it may not necessarily be sold as a package with the SOFTWARE. This LICENSE sets forth the terms and conditions of the SOFTWARE LICENSE only.
1. DEFINITIONS
1.1 Customer. Customer means the entity or individual that downloads the SOFTWARE.
2. GRANT OF LICENSE
2.1 Rights and Limitations of Grant. NVIDIA hereby grants Customer the following non-exclusive, non-transferable right to use the SOFTWARE, with the following limitations:
2.1.1 Rights. Customer may install and use one copy of the SOFTWARE on a single computer, and except for making one back-up copy of the Software, may not otherwise copy the SOFTWARE. This LICENSE of SOFTWARE may not be shared or used concurrently on different computers.
2.1.2 Linux/FreeBSD Exception. Notwithstanding the foregoing terms of Section 2.1.1, SOFTWARE designed exclusively for use on the Linux or FreeBSD operating systems, or other operating systems derived from the source code to these operating systems, may be copied and redistributed, provided that the binary files thereof are not modified in any way (except for unzipping of compressed files).
This is interesting I think.
Todd Zullinger wrote:
Would you moan at redhat if you couldn't get Oracle to work on it, yet Oracle claimed linux compatibility?
Of course. There's not much excuse for screwing up interfaces that an OS is supposed to provide.
Fedora's goal is to stay as close to upstream as possible and provide very current versions.
That might have made sense on the 'stable' branch of the kernel when there was an unstable branch for bizarre new ideas - like there was when the fedora project was started.
For the kernel, this means that the interface does change sometimes.
That's why major revision numbers were invented.
It is a simple fact. If you don't like it, please take your case to LKML. It is ridiculous, pointless, counterproductive, and IMO off-topic to bring it up here -- repeatedly.
If the kernel team insists on pushing out untested, broken stuff (and I think that's a given, with no development branch), it is now a distribution issue to decide what is fit to push out to the users. It is not off-topic at all, and I'll keep bringing it up as long as I have any hope that the fedora team cares at all if the update kernels they push out will boot. And if they don't care they should provide some truth-in-advertising and say exactly that on their project page.
on 10/29/2007 11:46 AM, Les Mikesell wrote:
David Boles wrote:
Just curious. The firewire drive problem. Are you staing that what you say, that they don't work well, is *all* Fedora and that they do work well with other distributions? Or that they don't work with the kernels, same number, of all of them?
In fedora, whether they work or not changes as fast as the weather in Chicago. Other distributions don't push changes out to users just to see what happens. In more sensible times, there was an explicit 'unstable' branch for kernel development so it was clear what had a chance of working in a distribution. Now all the wild and crazy changes are thrown in the same pot and it is left up to each distribution to guess what is ready for prime time.
Honestly if these drives were as important to me as you say that they are for you and that they worked with a different distro I would have switched a long, long time ago.
I did - and the Centosplus kernels have not given me any unpleasant surprises.
Interesting. CentOS 5 was supposed to be based on Fedora Core 6 work so now I wonder just what the difference(s) are?
As for Fedora itself. Since you no longer use it why to you care anymore?
Somebody in the thread at some point said:
Andy Green wrote:
You never had a closed source driver with a bug in? There's nothing for you to do but make a bug report and wait.
And that's what I would do with open source as well. The difference is that closed source OS's rarely change their driver interfaces, so it would be extremely unusual for something that already works and I have put into production to suddenly fail due to an update. In fedora, this has been a regular occurrence.
Don't really agree with that, I seem to recall a ton of security apps blew up with XP SP2. I notice that the argument forked there, suddenly we're on the perennial LM stable ABI rant when what led to it was options when faced with a buggy driver.
Good luck with the closed stuff.
-Andy
On Mon, 2007-10-29 at 11:55 -0600, Frank Cox wrote:
On Mon, 29 Oct 2007 13:23:06 -0400 Matthew Saltzman mjs@CLEMSON.EDU wrote:
With all due respect, sir, you don't know jack about my requirements, limitations, and objectives, or what environment I work in, or what tools are available in my field, or what I may or may not be in a position to do about it.
You could always try open hostility. It's highly recommended as a great way to get others to assist you with your problems.
This thread isn't about assistance. It's not even about rational discourse anymore.
I'm sorry, I assumed you were being deliberately provocative. I couldn't think of any other rationale for your remarks. Were you seriously suggesting that because I happen to have hardware that needs drivers that are only available in proprietary form, or that I happen to need Maple or Matlab or other commercial software with no OSS equivalent, or that my employer uses Exchange, that I have no business using Linux?
on 10/29/2007 11:46 AM, Les Mikesell wrote:
David Boles wrote:
Just curious. The firewire drive problem. Are you staing that what you say, that they don't work well, is *all* Fedora and that they do work well with other distributions? Or that they don't work with the kernels, same number, of all of them?
In fedora, whether they work or not changes as fast as the weather in Chicago. Other distributions don't push changes out to users just to see what happens. In more sensible times, there was an explicit 'unstable' branch for kernel development so it was clear what had a chance of working in a distribution. Now all the wild and crazy changes are thrown in the same pot and it is left up to each distribution to guess what is ready for prime time.
Honestly if these drives were as important to me as you say that they are for you and that they worked with a different distro I would have switched a long, long time ago.
I did - and the Centosplus kernels have not given me any unpleasant surprises.
Interesting. CentOS 5 was supposed ti be based on Fedora Core 6 work so now I wonder just what the difference(s) are?
As for Fedora itself. Since you no longer use it why to you care anymore?
On Mon, 29 Oct 2007 15:40:27 -0400 Matthew Saltzman mjs@CLEMSON.EDU wrote:
I'm sorry, I assumed you were being deliberately provocative. I couldn't think of any other rationale for your remarks.
You have either not read, or have failed to understand my previous remarks.
Were you seriously suggesting that because I happen to have hardware that needs drivers that are only available in proprietary form, or that I happen to need Maple or Matlab or other commercial software with no OSS equivalent, or that my employer uses Exchange, that I have no business using Linux?
I suggest that when you are doing things like interfacing with Microsoft Exchange and other things like that, your life will be much easier if you stick with a Microsoft software stack. If you wish to move to a Free Software stack, your life will be much simpler if you move to Free Software for those things rather than trying to ram the "old way of doing things" into a completely new system.
Of course, it's up to you if you want to do things the hard way or the easy way; you are welcome to use proprietary software on any platform you wish.
It becomes difficult to sympathize with people who are not prepared to help themselves, though. "I want to to X." In order to do X, you require Y and Z. "But I don't want to get Y. Just make it work with Z and make it look exactly the same as it did before. I don't want to have to learn anything new, ever".
If that's the situation, then you don't want Free Software. You want another box containing exactly what you have on your desk right now. "But it doesn't work very well."
See the contradiction?
Craig White wrote:
On Mon, 2007-10-29 at 11:45 -0600, Karl Larsen wrote:
Robert P. J. Day wrote:
"Like its sibling Firefox, the beauty of the open source Thunderbird ^^^^^^^^^^^ email client is its extensibility. Sure we love our web-based email like Gmail, but Thunderbird is the ultimate open source desktop email ^^^^^^^^^^^ app."
in response to the above, i'm guessing karl will call me "stupid."
rday
Shame on you rday, your using a non-fedora email system. Say what now?
thunderbird is part of fedora base packages
Craig
I guess Karl feels that if it isn't the default, then it is "non-fedora". (Would that make KDE non-fedora as well?)
One thing I don't know - what is an "email system?" I know what a mail server is, and what an e-mail client is, but what is an "email system"? I have seen the term used for a server or servers that handle mail for a network, but that definition doesn't seam to apply here.
Mikkel
Frank Cox wrote:
On Mon, 29 Oct 2007 03:45:29 -0400 Ric Moore wayward4now@gmail.com wrote:
I really miss the old iron.
I don't have much experience with other CP/M emulators, but the CP/M mode of the Commodore 128 is a great deal of fun to play with through the Vice emulator.
You can get it here: http://viceteam.org/
The C128 was interesting in that it had 2 processors. It would switch to the Z80 processor when running in the CP/M mode. It also ran CP/M 3 or CP/M Plus (Different names for the same thing.) so you could use more then 64K of RAM using memory paging.
Mikkel
Frank Cox wrote:
On Mon, 29 Oct 2007 13:00:36 +0000 Andy Green andy@warmcat.com wrote:
Apparently the concern historically has been explained that they will lay themselves open to patent attacks by showing in the source what they are actually doing.
That sounds like a very odd explanation, if indeed that is the way that they described it. "I don't want to tell you what I do for a living because I would have to admit that I'm a burglar."
It seems to me that statement like that could provide "probable cause" (or whatever the correct term is) for a patent holder to get a court order to have those guys checked out for patent violations.
I think the difference is that if they do not release the source, then you have everybody guessing if it is their patent that is infringed on, but if they do release the source, then it is easier to tell what patents. Otherwise it is to big of a gamble - remember, you will probably get stuck with all the lawyers fees if you guess wrong about witch patent(s).
Mikkel
Frank Cox wrote:
I'm sorry, I assumed you were being deliberately provocative. I couldn't think of any other rationale for your remarks.
You have either not read, or have failed to understand my previous remarks.
Or just considered them wrong...
Were you seriously suggesting that because I happen to have hardware that needs drivers that are only available in proprietary form, or that I happen to need Maple or Matlab or other commercial software with no OSS equivalent, or that my employer uses Exchange, that I have no business using Linux?
I suggest that when you are doing things like interfacing with Microsoft Exchange and other things like that, your life will be much easier if you stick with a Microsoft software stack. If you wish to move to a Free Software stack, your life will be much simpler if you move to Free Software for those things rather than trying to ram the "old way of doing things" into a completely new system.
Sendmail is the "old way of doing things" and it's completely agnostic as to whether it runs on what you consider Free software or any commercial version of unix.
Of course, it's up to you if you want to do things the hard way or the easy way; you are welcome to use proprietary software on any platform you wish.
Standards compliance is what makes things interoperate. There is next to no relationship to whether or not the components are proprietary or not.
It becomes difficult to sympathize with people who are not prepared to help themselves, though. "I want to to X." In order to do X, you require Y and Z. "But I don't want to get Y. Just make it work with Z and make it look exactly the same as it did before. I don't want to have to learn anything new, ever".
That makes no sense at all. We want X _and_ Y to work together.
If that's the situation, then you don't want Free Software. You want another box containing exactly what you have on your desk right now. "But it doesn't work very well."
See the contradiction?
No. The contradiction is the part that calls itself free yet tries to dictate what other components can do.
Karl Larsen wrote:
You might have a nvidia video card on your motherboard. There are two choices here. Try to use the nvidia or turn it off and plug in your old known video card. Today I wish I had done the latter because using nvidia with f7 is a pain.
I really do not see a new Linux user ever getting his/her computer working with nvidia. You need to go to the nvidia web page and get a tarball and install it, not a new person's thing, or you can get 4 rpm files and learn to use --nodeps at the proper time.
Hoping that the updates would by now have some nvidia help, after getting 236 updates last night on my f7-64 bit system it did not fix the problem. I used the 4 rpm files from www.atrpms.net which worked but maybe not well. I heard from Ric Moore that the tarball is the way to go. I will try that on f8.
A bug I keep forgetting to file is the following. A really bad problem with nvidia is the missing pointer when X windows boots up. You can do nothing! This is fixed by edit of the /etc/X11/xrog.conf file adding you want to use a software pointer.
But this will not work if grub.conf has a kernel directive to use rhxxx which hides the boot up output. While that standard kernel directive exists you can not get a pointer period.
This bug makes f7 and I expect f8 useless to a new user with nvidia.
I have read through the thread and decided that a single response is better than a dozen little responses.
I have used Nvidia since almost 2 months of fighting with ATI's driver a few years ago. I had no issues with Nvidia until I installed F7. Well, it wasn't the install but a kernel upgrade later when the problems started. I tried the Livna, Freshrpms and Nvidia versions of the same driver. All caused the same issues. None worked.
I knew my card was slower so I changed it. Surprise, the drivers worked. I now use the freshrpms due to dkms support. Better than having to re-make the Nvidia drivers or remove and upgrade the Livna drivers.
Now I don't put the blame on Nvidia as many in this thread have. I don't blame Fedora/RedHat, I blame the US government for allowing Software patents. This is where the issue stems from. No software patents, then no patent issues. Of course there is still copyright issues but that is another matter.
AMD/ATI merger is going to be good for OSS, as long as AMD releases decent drivers. Now the other end of this is it may force Nvidia to release OSS drivers.
While I was having problems, and the problem extended to the nv driver as well, I submitted bugs to both Nvidia and RedHat. At least Nvidia took the time to respond with some suggestions. RedHat just told me to take a hike as they had nothing to do with the closed source drivers. Even after submitting stack traces of the issues for the nv driver that were the same as the binary driver.
FWIW, in some recent kernels, I have had the same issue with an ATI card on FC6 that I have under F7. System freezes and Xorg running 95% of the processor. I don't point my finger to Nvidia but to the Kernel developers that have made a change. On the Nvidia forums for the problem (freezing) that I had, someone installed a custom kernel and fixed all their driver problems.
I do agree that if we want to get people converted from Windows to Linux, we have to get it working out of the box, not with two days of searching Google and hopefully fixing their issues.
I have two working machines that use Nvidia graphics and no issues lately.
From what I understand, the Nvidia driver requires more changes than just the addition of a module to the OS. I understand it changes some other files if you use the binary blob. I used the Livna rpms for ages but in F7, I find that the freshrpms is much nicer. No removing and re-installing the driver.
I think that the Fedora staff have to be willing to look at a closed source driver issue and submit a bug report up-stream in behalf of their users.
As for video card issues. I know a Windows user that last year dumped all ATI video cards due to driver issues. And Vista had major headaches with drivers. So the driver issue isn't just Linux. In the Windows world, Microsoft will throw money at the problem.
Maybe we can ask this group to get involved.
Know Any Hardware Needing Better Linux Support? http://linux.slashdot.org/article.pl?sid=07/10/28/0233231
On Mon, 29 Oct 2007 15:19:59 -0500 "Mikkel L. Ellertson" mikkel@infinity-ltd.com wrote:
The C128 was interesting in that it had 2 processors. It would switch to the Z80 processor when running in the CP/M mode. It also ran CP/M 3 or CP/M Plus (Different names for the same thing.) so you could use more then 64K of RAM using memory paging.
It's my understanding that the CP/M mode of the C128 was very under-utilized and under-appreciated.
I had a rather good accounting program that just happened to be a full-fledged database program as well. I thought it was called "TAS: The Accounting Solution", but I just looked that up and there still is a "TAS: The Accounting Solution" and all they mentioned there was DOS so it might have been something else. It was definitely CP/M; I remember that it came on Kaypro formatted disks and the Commodore 1571 drive read them with no problem. I also learned dBase II on the C128 in CP/M mode.
I had a fairly decent little collection of CP/M software for my C128, at one time....
There is on the Nvidia (put in Google) site a newest version of of there drivers for Linux. You have to d/l a "run" file which you select from the first questions on the page. Then I discovered it will not run if the X server is running.
Just found out to turn off the X server in a root Terminal type init 3. There you go to the directory with the run file and as root type sh NVID..... and it will start. On my computer it could not find a ready made kernel addition so it made on on the spot and then went on to do other things.
Then I typed init 5 and here I am back on X windows with MUCH sharper pictures and things and the pointer is new.
This is how you should get your Nvidia on Linux if you want the best qualities of Nvidia color.
Les Mikesell wrote:
In my (limited) experience closed drivers create serious problems, as you never know if they will work with newer kernels or other software, so you cannot rely on them too much.
In my experience, open source drivers have as many or more problems than the closed source versions.
Well, virtually all the open source drivers I use are maintained by someone who updates them for new versions of Linux and/or Fedora. By contrast, the few closed source drivers I have used usually state something like "Compatible with Linux 2.2.10" and when they don't work with Linux 2.6.10 there is nothing you can do about it. (I came across this recently with a driver from Avaya for an Orinoco WiFi card.)
_If_ you are Alan Cox or some number of hackers with equivalently specialized skills that you could probably count on one hand, having the source code available might be of some value when the supplied binary doesn't work. The rest of us report the bug and wait, and again in my experience over the last couple of decades, the closed source providers are at least equally responsive in this scenario.
Really? When did you last get a positive response from a closed source provider?
Frank Cox wrote:
On Mon, 29 Oct 2007 15:19:59 -0500 "Mikkel L. Ellertson" mikkel@infinity-ltd.com wrote:
The C128 was interesting in that it had 2 processors. It would switch to the Z80 processor when running in the CP/M mode. It also ran CP/M 3 or CP/M Plus (Different names for the same thing.) so you could use more then 64K of RAM using memory paging.
It's my understanding that the CP/M mode of the C128 was very under-utilized and under-appreciated.
I had a rather good accounting program that just happened to be a full-fledged database program as well. I thought it was called "TAS: The Accounting Solution", but I just looked that up and there still is a "TAS: The Accounting Solution" and all they mentioned there was DOS so it might have been something else. It was definitely CP/M; I remember that it came on Kaypro formatted disks and the Commodore 1571 drive read them with no problem. I also learned dBase II on the C128 in CP/M mode.
I had a fairly decent little collection of CP/M software for my C128, at one time....
I had only limited experience with the 128 - A friend had one. I was on my third CP/M machine myself - this one had a 20M hard drive, as well as both standard and double-density 5-1/4" drives. We found that IBM-PC format made the best exchange format between us.
I still have a fair collection of software for CP/M - including Word Star word processor, Personal Perl database, Both standard and macro assemblers, a C compiler, modem programs, etc. I wish I could legally add them to one of the CP/M software sites.
I used to write a lot of serial port and modem overlay files for the modem program. (MEX?) Some were simple changes to add the carrier detect bit, while others were full overlays for new hardware based on the default template. I don't know if they are still floating around. There should also be a Battle Ship type game and a Number Squares game written in C out there someplace. (Text based.)
Mikkel
Les Mikesell wrote:
The difference is that closed source OS's rarely change their driver interfaces, so it would be extremely unusual for something that already works and I have put into production to suddenly fail due to an update.
I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
In fedora, this has been a regular occurrence.
Such as?
Timothy Murphy wrote:
_If_ you are Alan Cox or some number of hackers with equivalently specialized skills that you could probably count on one hand, having the source code available might be of some value when the supplied binary doesn't work. The rest of us report the bug and wait, and again in my experience over the last couple of decades, the closed source providers are at least equally responsive in this scenario.
Really? When did you last get a positive response from a closed source provider?
In the windows/mac worlds, I can't say that I've needed anything in years that wasn't already done and available on the vendor's web site.
Timothy Murphy wrote:
The difference is that closed source OS's rarely change their driver interfaces, so it would be extremely unusual for something that already works and I have put into production to suddenly fail due to an update.
I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
I think you misunderstood. I said closed source - as opposed to Linux - doesn't change driver interfaces often. With Linux the kernel changes continuously but it is up to the distribution what is shipped. RHEL maintains something stable. Fedora doesn't.
In fedora, this has been a regular occurrence.
Such as?
Firewire disk drives are my favorite example. The MPT scsi controllers on IBM servers and some Dells have also failed to work with several fedora kernels. And these are all open source that can't even stay compatible with itself. I know better than to waste my time using fedora with anything that isn't open source - even vmware.
David Boles wrote:
Honestly if these drives were as important to me as you say that they are for you and that they worked with a different distro I would have switched a long, long time ago.
I did - and the Centosplus kernels have not given me any unpleasant surprises.
Interesting. CentOS 5 was supposed ti be based on Fedora Core 6 work so now I wonder just what the difference(s) are?
The difference is that RHEL (the sources that Centos recompiles) cares enough about their user base to maintain a stable kernel interface for the life of the distribution version - which is much longer than fedora's. The down side is that applications don't get version upgrades as the distro version ages, just security and bugfix updates.
As for Fedora itself. Since you no longer use it why to you care anymore?
I have more than one machine, so I track fedora on a box that I can afford to have fail to boot on occasion to know what to expect from the next RHEL/Centos.
On Mon, 2007-10-29 at 13:49 -0600, Frank Cox wrote:
On Mon, 29 Oct 2007 15:40:27 -0400 Matthew Saltzman mjs@CLEMSON.EDU wrote:
I'm sorry, I assumed you were being deliberately provocative. I couldn't think of any other rationale for your remarks.
You have either not read, or have failed to understand my previous remarks.
I read this:
If you depend on proprietary software, then Linux is not for you. Microsoft and/or Apple products are a much better fit for your needs.
and this:
There are all kinds of licenses. If your requirements include "proprietary, closed-source and immutable software", then Linux is not for you.
I'm sorry if this is the first time you have heard this.
the entirety of two back-to-back messages. The message seemed clear enough.
Were you seriously suggesting that because I happen to have hardware that needs drivers that are only available in proprietary form, or that I happen to need Maple or Matlab or other commercial software with no OSS equivalent, or that my employer uses Exchange, that I have no business using Linux?
I suggest that when you are doing things like interfacing with Microsoft Exchange and other things like that, your life will be much easier if you stick with a Microsoft software stack. If you wish to move to a Free Software stack, your life will be much simpler if you move to Free Software for those things rather than trying to ram the "old way of doing things" into a completely new system.
Of course, it's up to you if you want to do things the hard way or the easy way; you are welcome to use proprietary software on any platform you wish.
It becomes difficult to sympathize with people who are not prepared to help themselves, though. "I want to to X." In order to do X, you require Y and Z. "But I don't want to get Y. Just make it work with Z and make it look exactly the same as it did before. I don't want to have to learn anything new, ever".
If that's the situation, then you don't want Free Software. You want another box containing exactly what you have on your desk right now. "But it doesn't work very well."
I support efforts to get hardware manufacturers to open their drivers and software vendors to open their standards. I support the efforts of the Samba team and the Evolution Exchange Connector team and the NTFS team to make open-source tools that interoperate with those proprietary systems. I've done--and continue to do--my share of living with bugs and reporting bugs in those tools (and lobbying vendors) so I can work in an open-source environment. I appreciate the difficulties that those developers face.
I develop open-source tools (including tools that interoperate with proprietary software) as part of my work and I advocate for open-source tools to my colleagues. I've contributed patches and diagnostics to a few Red Hat packages. But I think the world is a big place with room for lots of business models. Open source is a superior model in many ways, but not superior (even if no worse than proprietary) in others. Proprietary software isn't going to disappear tomorrow. While we are working on vendors to make their products more open-source friendly, I still need to use what's available to get my work done.
And I resent being told that because of circumstances I can't control, I'm somehow not worthy of being a member of the Linux community. Your quarrel is with the vendors. Picking fights with your users is counterproductive.
See the contradiction?
Les Mikesell wrote:
David Boles wrote:
Honestly if these drives were as important to me as you say that they are for you and that they worked with a different distro I would have switched a long, long time ago.
I did - and the Centosplus kernels have not given me any unpleasant surprises.
Interesting. CentOS 5 was supposed ti be based on Fedora Core 6 work so now I wonder just what the difference(s) are?
The difference is that RHEL (the sources that Centos recompiles) cares enough about their user base to maintain a stable kernel interface for the life of the distribution version - which is much longer than fedora's. The down side is that applications don't get version upgrades as the distro version ages, just security and bugfix updates.
That depends on the applications themselves. Some of the desktop applications in particular do get version upgrades too.
Rahul
Scott van Looy wrote:
The point Les appears to be missing is that the binary blob doesn't stay the same because the cards it supports don't stay the same. And as NVidia upgrade their hardware, they'll slowly stop supporting the older hardware.
I'm not convinced that is a correct or valid statement.
I'm running a system with a TNT2 card from NVidia that I bought sometime in 2000 or 2001. This hardware has not be produced in a long time. The drivers are still available and fixes are made in the "NVidia Legacy GPU Channel".
Les Mikesell wrote:
If the kernel team insists on pushing out untested, broken stuff (and I think that's a given, with no development branch), it is now a distribution issue to decide what is fit to push out to the users. It is not off-topic at all, and I'll keep bringing it up as long as I have any hope that the fedora team cares at all if the update kernels they push out will boot. And if they don't care they should provide some truth-in-advertising and say exactly that on their project page.
This list isn't the best way to reach the kernel maintainers. They have a list just for those discussions (fedora-kernel-list). If you're serious about proposing changes that you feel are needed, then that's a far better place than here to do it. Otherwise, you're just pissing into the wind (and causing it to blow back on the rest of us).
Please be sure to suggest to Dave Jones and the others on the kernel list that you're not sure if they care if kernels they push to the stable update repo are bootable. I'm sure they'll all appreciate that.
Thus, Les Mikesell at Mon Oct 29 22:07:52 2007 inscribed:
I think you misunderstood. I said closed source - as opposed to Linux - doesn't change driver interfaces often. With Linux the kernel changes continuously but it is up to the distribution what is shipped. RHEL maintains something stable. Fedora doesn't.
The support and the ABI/API stability is part of what you pay for with your subscription with RHEL. You pay nothing for Fedora. Stop trying to turn Fedora in to RHEL or CentOS, they have different purposes and goals.
Rahul Sundaram wrote:
The difference is that RHEL (the sources that Centos recompiles) cares enough about their user base to maintain a stable kernel interface for the life of the distribution version - which is much longer than fedora's. The down side is that applications don't get version upgrades as the distro version ages, just security and bugfix updates.
That depends on the applications themselves. Some of the desktop applications in particular do get version upgrades too.
OK, I suppose it's possible. But the mid-life 4.x distro is still providing firefox 1.5.x, subversion 1.1.x, evolution 2.0.x. Are those things you'd want on your desktop much longer?
Les Mikesell wrote:
Rahul Sundaram wrote:
The difference is that RHEL (the sources that Centos recompiles) cares enough about their user base to maintain a stable kernel interface for the life of the distribution version - which is much longer than fedora's. The down side is that applications don't get version upgrades as the distro version ages, just security and bugfix updates.
That depends on the applications themselves. Some of the desktop applications in particular do get version upgrades too.
OK, I suppose it's possible.
It is not merely possible. It happens all the time.
But the mid-life 4.x distro is still
providing firefox 1.5.x, subversion 1.1.x, evolution 2.0.x. Are those things you'd want on your desktop much longer?
That depends on what you want out of the desktop. If you want the latest and greatest at all times, Fedora will serve that need. If you prefer targeted fixes, commercial support etc, RHEL will meet that need better.
Rahul
on 10/29/2007 6:20 PM, Les Mikesell wrote:
David Boles wrote:
Honestly if these drives were as important to me as you say that they are for you and that they worked with a different distro I would have switched a long, long time ago.
I did - and the Centosplus kernels have not given me any unpleasant surprises.
Interesting. CentOS 5 was supposed ti be based on Fedora Core 6 work so now I wonder just what the difference(s) are?
The difference is that RHEL (the sources that Centos recompiles) cares enough about their user base to maintain a stable kernel interface for the life of the distribution version - which is much longer than fedora's. The down side is that applications don't get version upgrades as the distro version ages, just security and bugfix updates.
As for Fedora itself. Since you no longer use it why to you care anymore?
I have more than one machine, so I track fedora on a box that I can afford to have fail to boot on occasion to know what to expect from the next RHEL/Centos.
Ah! I see now. A 'bitch about it' box just for Fedora. ;-)
David Boles wrote:
As for Fedora itself. Since you no longer use it why to you care anymore?
I have more than one machine, so I track fedora on a box that I can afford to have fail to boot on occasion to know what to expect from the next RHEL/Centos.
Ah! I see now. A 'bitch about it' box just for Fedora. ;-)
I look at it as having 3 chances to get the broken stuff fixed before I'm going to have to live with it in Centos... But, I'd give it much more of a workout if wasn't such a pain to keep updated under vmware since the interface to vmware tools changes all the time.
Rahul Sundaram wrote:
That depends on the applications themselves. Some of the desktop applications in particular do get version upgrades too.
OK, I suppose it's possible.
It is not merely possible. It happens all the time.
But the mid-life 4.x distro is still
providing firefox 1.5.x, subversion 1.1.x, evolution 2.0.x. Are those things you'd want on your desktop much longer?
That depends on what you want out of the desktop. If you want the latest and greatest at all times, Fedora will serve that need. If you prefer targeted fixes, commercial support etc, RHEL will meet that need better.
I want something tested but not ancient. Neither disto provides that except for the first few months after an RHEL cut. But I'm usually happy with old kernel and server apps (except subversion and dovecot...). It's generally just the desktop stuff that changes fast enough to care about.
On Mon, 2007-10-29 at 18:10 -0500, Les Mikesell wrote:
Rahul Sundaram wrote:
That depends on the applications themselves. Some of the desktop applications in particular do get version upgrades too.
OK, I suppose it's possible.
It is not merely possible. It happens all the time.
But the mid-life 4.x distro is still
providing firefox 1.5.x, subversion 1.1.x, evolution 2.0.x. Are those things you'd want on your desktop much longer?
That depends on what you want out of the desktop. If you want the latest and greatest at all times, Fedora will serve that need. If you prefer targeted fixes, commercial support etc, RHEL will meet that need better.
I want something tested but not ancient. Neither disto provides that except for the first few months after an RHEL cut. But I'm usually happy with old kernel and server apps (except subversion and dovecot...). It's generally just the desktop stuff that changes fast enough to care about.
---- the problem is and will always be the interdependence among the various packages. While it may be convenient to look at things in a vacuum, they simply don't work that way - one package update requires updates on requisite packages which impacts something else down the line. The very thing that makes you rich, makes you poor.
On Windows, software is self referenced so that there isn't the package dependencies that you have with Linux. But that's also why each text editor/word processor etc. doesn't have to include a dictionary on Linux and why they all include a dictionary on Windows.
Craig
on 10/29/2007 7:03 PM, Les Mikesell wrote:
David Boles wrote:
As for Fedora itself. Since you no longer use it why to you care anymore?
I have more than one machine, so I track fedora on a box that I can afford to have fail to boot on occasion to know what to expect from the next RHEL/Centos.
Ah! I see now. A 'bitch about it' box just for Fedora. ;-)
I look at it as having 3 chances to get the broken stuff fixed before I'm going to have to live with it in Centos... But, I'd give it much more of a workout if wasn't such a pain to keep updated under vmware since the interface to vmware tools changes all the time.
I thought that they, VMware, fixed that with 6.0x.
Craig White wrote:
I want something tested but not ancient. Neither disto provides that except for the first few months after an RHEL cut. But I'm usually happy with old kernel and server apps (except subversion and dovecot...). It's generally just the desktop stuff that changes fast enough to care about.
the problem is and will always be the interdependence among the various packages. While it may be convenient to look at things in a vacuum, they simply don't work that way - one package update requires updates on requisite packages which impacts something else down the line. The very thing that makes you rich, makes you poor.
That's a possibility, but rarely the case except perhaps for the Gnome/KDE environments themselves. Usually it is possible to build current packages on older RHEL versions, but then you have to maintain them yourself.
On Windows, software is self referenced so that there isn't the package dependencies that you have with Linux. But that's also why each text editor/word processor etc. doesn't have to include a dictionary on Linux and why they all include a dictionary on Windows.
I'm not sure what you run under windows, but my email and word processor share a dictionary and I believe they document the interface to it.
Les Mikesell wrote:
The difference is that closed source OS's rarely change their driver interfaces, so it would be extremely unusual for something that already works and I have put into production to suddenly fail due to an update.
I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
I think you misunderstood. I said closed source - as opposed to Linux - doesn't change driver interfaces often. With Linux the kernel changes continuously but it is up to the distribution what is shipped. RHEL maintains something stable. Fedora doesn't.
But a closed source driver has to interface with the Linux kernel.
In fedora, this has been a regular occurrence.
Such as?
Firewire disk drives are my favorite example. The MPT scsi controllers on IBM servers and some Dells have also failed to work with several fedora kernels.
Has this actually got anything to do with Fedora? Did Fedora modify the kernel SCSI driver in some way? Or is the driver not in the kernel?
Just to repeat my argument. I would have thought a closed source driver was much less likely to keep up with kernel changes than a Linux driver. My (limited) experience with closed source Linux drivers is that they generally specify that they work with a given kernel.
On Mon, 2007-10-29 at 10:20 -0500, Les Mikesell wrote:
One piece of software does not get to dictate what others you can run regardless of the unrealistic fanaticism involved.
Please explain that to Bill Gates...
On Mon, 2007-10-29 at 15:18 +0000, Scott van Looy wrote:
If you want to run linux, you do some investigation online. Google is your friend. It's pretty easy to set up the nvidia drivers, the nv drivers are included out of the box so it'll all work anyway, the only thing that you won't get is 3d acceleration.
There can be a few other things that don't work, such as being able to hibernate your box with the nv drivers... But they mayn't be important to you (that's "you" in general, not you in particular).
I discovered that little annoyance with my Laptop. And there's other things, like the video modes offered.
On Mon, 2007-10-29 at 15:31 -0600, Karl Larsen wrote:
This is how you should get your Nvidia on Linux if you want the best qualities of Nvidia color.
Correction: This is ONE way that you CAN set up your NVidia card.
For some, the drivers from NVidia caused problems. Go googling for why not to use the binary drivers from NVidia, and you'll see the explanation of how they stuffed up standard files to suit themselves.
Tim wrote:
On Mon, 2007-10-29 at 10:20 -0500, Les Mikesell wrote:
One piece of software does not get to dictate what others you can run regardless of the unrealistic fanaticism involved.
Please explain that to Bill Gates...
I'm not aware of any Microsoft restrictions against combining components as long as the end user has met the license requirements individually for each of them. And I'd guess any attempt to do so would be met with lawsuits over anticompetitive behavior.
Timothy Murphy wrote:
The difference is that closed source OS's rarely change their driver interfaces, so it would be extremely unusual for something that already works and I have put into production to suddenly fail due to an update.
I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
I think you misunderstood. I said closed source - as opposed to Linux - doesn't change driver interfaces often. With Linux the kernel changes continuously but it is up to the distribution what is shipped. RHEL maintains something stable. Fedora doesn't.
But a closed source driver has to interface with the Linux kernel.
Most of my closed source drivers are used with closed source OS's and are simply not an issue. The only one I use regularly with Linux is the vmware module which is moderated by having an interface of its own that can be recompiled if needed.
In fedora, this has been a regular occurrence.
Such as?
Firewire disk drives are my favorite example. The MPT scsi controllers on IBM servers and some Dells have also failed to work with several fedora kernels.
Has this actually got anything to do with Fedora? Did Fedora modify the kernel SCSI driver in some way? Or is the driver not in the kernel?
Distributions generally don't ship every kernel exactly as blessed by Linus. Fedora chose to ship things that didn't work. It didn't really matter to me if it was a fedora-local change or they just passed it through - I think some of each has happened.
Just to repeat my argument. I would have thought a closed source driver was much less likely to keep up with kernel changes than a Linux driver. My (limited) experience with closed source Linux drivers is that they generally specify that they work with a given kernel.
Things work through the 7-year life of an RHEL (and thus Centos) kernel. They sometimes work through the 2 week (or so) life of a fedora kernel. It's a strange model where you get the software for free but pay them (in the RHEL case) not to break it with updates.
David Boles wrote:
on 10/29/2007 7:03 PM, Les Mikesell wrote:
David Boles wrote:
As for Fedora itself. Since you no longer use it why to you care anymore?
I have more than one machine, so I track fedora on a box that I can afford to have fail to boot on occasion to know what to expect from the next RHEL/Centos.
Ah! I see now. A 'bitch about it' box just for Fedora. ;-)
I look at it as having 3 chances to get the broken stuff fixed before I'm going to have to live with it in Centos... But, I'd give it much more of a workout if wasn't such a pain to keep updated under vmware since the interface to vmware tools changes all the time.
I thought that they, VMware, fixed that with 6.0x.
Version 6.0 of what? I don't see how they can fix it permanently without a sane interface on the Linux side to interoperate with.
On Mon, 29 Oct 2007, Karl Larsen wrote:
There is on the Nvidia (put in Google) site a newest version of of there drivers for Linux. You have to d/l a "run" file which you select from the first questions on the page. Then I discovered it will not run if the X server is running.
Just found out to turn off the X server in a root Terminal type init 3. There you go to the directory with the run file and as root type sh NVID..... and it will start. On my computer it could not find a ready made kernel addition so it made on on the spot and then went on to do other things.
Then I typed init 5 and here I am back on X windows with MUCH sharper pictures and things and the pointer is new.
This is how you should get your Nvidia on Linux if you want the best qualities of Nvidia color.
i've finally broken the code of the karl-bot. it's a simple, repeating pattern, as follows:
1) karl posts, asking a question, accompanied by breathtakingly inane initial suggestion as to the solution:
"i need to switch to runlevel 3 -- i guess it's time for my rescue CD"
2) numerous mailing list members try to save karl from himself and respond to him, attempting to educate him
3) from all these responses, karl puts together a tutorial in which he gives no credit whatsoever to any of the people who tried to help him, and presents his work as if he is now an authority (see above).
4) at no time does it dawn on karl that the helpful tutorial he has just posted has been available on the net the whole time, like here, for instance:
http://forum.compiz-fusion.org/showthread.php?t=860
where it is not only correct, but is presented in something resembling syntactially proper english.
... and the life cycle of the karl-bot continues.
rday
Dogbert: "Dogbert Technical Support, how may I help you?" User: "I'm having a problem with -- Dogbert: "Shut up and reboot." User: "Hey, it worked! Thank -- Dogbert: "Shut up and hang up." <click> Dogbert: "My average call time is improving."
On Mon, 29 Oct 2007, Timothy Murphy wrote:
Les Mikesell wrote:
The difference is that closed source OS's rarely change their driver interfaces, so it would be extremely unusual for something that already works and I have put into production to suddenly fail due to an update.
I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
um ... no. the kernel *internal* interfaces may change, but the interface that is presented to the outside world is very stable.
http://lxr.linux.no/source/Documentation/stable_api_nonsense.txt
rday
Robert P. J. Day wrote:
The difference is that closed source OS's rarely change their driver interfaces, so it would be extremely unusual for something that already works and I have put into production to suddenly fail due to an update.
I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
um ... no. the kernel *internal* interfaces may change, but the interface that is presented to the outside world is very stable.
http://lxr.linux.no/source/Documentation/stable_api_nonsense.txt
That's fine if you believe that the linux kernel will always include every device driver, filesystem, and feature you'll ever want. Don't ask me to join you in that leap of faith (or arrogance, or whatever it is that makes you think interoperability is unnecessary). Even if deprive yourself of devices not handled by Linux drivers, vmware is just too handy to avoid.
On Tue, 30 Oct 2007, Les Mikesell wrote:
Robert P. J. Day wrote:
The difference is that closed source OS's rarely change their driver interfaces, so it would be extremely unusual for something that already works and I have put into production to suddenly fail due to an update.
I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
um ... no. the kernel *internal* interfaces may change, but the interface that is presented to the outside world is very stable.
http://lxr.linux.no/source/Documentation/stable_api_nonsense.txt
That's fine if you believe that the linux kernel will always include every device driver, filesystem, and feature you'll ever want. Don't ask me to join you in that leap of faith (or arrogance, or whatever it is that makes you think interoperability is unnecessary)...
what are you babbling about, les? all i did was clarify the distinction between the stability of the *internal* kernel interface and that of the *external* kernel interface. i said nothing whatsoever about driver support or interoperability.
if i were you, i'd switch to decaf.
rday
Robert P. J. Day wrote:
On Tue, 30 Oct 2007, Les Mikesell wrote:
Robert P. J. Day wrote:
The difference is that closed source OS's rarely change their driver interfaces, so it would be extremely unusual for something that already works and I have put into production to suddenly fail due to an update.
I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
um ... no. the kernel *internal* interfaces may change, but the interface that is presented to the outside world is very stable.
http://lxr.linux.no/source/Documentation/stable_api_nonsense.txt
That's fine if you believe that the linux kernel will always include every device driver, filesystem, and feature you'll ever want. Don't ask me to join you in that leap of faith (or arrogance, or whatever it is that makes you think interoperability is unnecessary)...
what are you babbling about, les? all i did was clarify the distinction between the stability of the *internal* kernel interface and that of the *external* kernel interface. i said nothing whatsoever about driver support or interoperability.
Errr, what? How can you say that the internal kernel interface does not relate to the drivers that use them?
Thus, Les Mikesell at Mon Oct 29 23:45:49 2007 inscribed:
Craig White wrote:
I want something tested but not ancient. Neither disto provides that except for the first few months after an RHEL cut. But I'm usually happy with old kernel and server apps (except subversion and dovecot...). It's generally just the desktop stuff that changes fast enough to care about.
the problem is and will always be the interdependence among the various packages. While it may be convenient to look at things in a vacuum, they simply don't work that way - one package update requires updates on requisite packages which impacts something else down the line. The very thing that makes you rich, makes you poor.
That's a possibility, but rarely the case except perhaps for the Gnome/KDE environments themselves. Usually it is possible to build current packages on older RHEL versions, but then you have to maintain them yourself.
Let me see if I have understood this right. You look at subversion or dovecot in RHEL5, and you can determine by just looking at the version number that they are insufficient for your needs, even if they were suitable five months earlier, when RHEL5 was just released?
That, to me at least, would indicate that you have totally failed to understand the purpose of RHEL and the value that it provides to Enterprise customers. It also would indicate that you do not understand how Red Hat maintain each release of RHEL (although you are in good company there, as some customers don't understand that either to start with).
On Tue, 30 Oct 2007, Les Mikesell wrote:
Robert P. J. Day wrote:
On Tue, 30 Oct 2007, Les Mikesell wrote:
Robert P. J. Day wrote:
The difference is that closed source OS's rarely change their driver interfaces, so it would be extremely unusual for something that already works and I have put into production to suddenly fail due to an update.
I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
um ... no. the kernel *internal* interfaces may change, but the interface that is presented to the outside world is very stable.
http://lxr.linux.no/source/Documentation/stable_api_nonsense.txt
That's fine if you believe that the linux kernel will always include every device driver, filesystem, and feature you'll ever want. Don't ask me to join you in that leap of faith (or arrogance, or whatever it is that makes you think interoperability is unnecessary)...
what are you babbling about, les? all i did was clarify the distinction between the stability of the *internal* kernel interface and that of the *external* kernel interface. i said nothing whatsoever about driver support or interoperability.
Errr, what? How can you say that the internal kernel interface does not relate to the drivers that use them?
errr ... did you actually *read* the document for which i provided a URL, les? ***of course*** changing the internel kernel interfaces will undoubtedly affect the drivers which have been written to those interfaces. and when those interfaces change, ***of course*** those drivers will have to be updated to conform to the new interfaces.
from the document above:
"Linux kernel development is continuous and at a rapid pace, never stopping to slow down. As such, the kernel developers find bugs in current interfaces, or figure out a better way to do things. If they do that, they then fix the current interfaces to work better. When they do so, function names may change, structures may grow or shrink, and function parameters may be reworked. If this happens, all of the instances of where this interface is used within the kernel are fixed up at the same time, ensuring that everything continues to work properly."
see how that works, les? when enough people in the kernel development community decide that an internal interface needs to change, everyone (theoretically) works together to make that transition as seamless and trouble-free as possible. and (note well) that all of that should be invisible to the outside world -- only driver developers need care. so what's your problem? what exactly bothers you about this process?
rday
p.s. also note, les, that if a given driver is in-tree, that driver's author doesn't even need to know about this, since it's the duty of those people making the change to *also* take care of all code that calls that interface. did you seriously think that someone might change an internal kernel interface and just leave all the broken calls to it hanging around in the kernel source tree?
Today Ed Greshko did spake thusly:
Scott van Looy wrote:
The point Les appears to be missing is that the binary blob doesn't stay the same because the cards it supports don't stay the same. And as NVidia upgrade their hardware, they'll slowly stop supporting the older hardware.
I'm not convinced that is a correct or valid statement.
I'm running a system with a TNT2 card from NVidia that I bought sometime in 2000 or 2001. This hardware has not be produced in a long time. The drivers are still available and fixes are made in the "NVidia Legacy GPU Channel".
They recently split the drivers in two. Your driver will be getting security fixes and some bug fixes and no more. The goal for them will be to create something stable that they don't need to touch any more. Makes business sense, no?
It's up to them what they do. There's no way someone else can easily step in if they decide not to continue actively maintaining old hardware.
Just because this old card works for you at the moment doesn't mean it always will. NVidia are very good at supporting legacy hardware, other companies are less so, I was merely using them as an example as that's what we were talking about.
Scott van Looy wrote:
Just because this old card works for you at the moment doesn't mean it always will. NVidia are very good at supporting legacy hardware, other companies are less so, I was merely using them as an example as that's what we were talking about.
If/when it stops working I'm sure it will be due to the hardware crapping out.
Yes, NVidia is very good at supporting their legacy hardware...so it would have been appropriate to use a company that isn't, just to be fair and not have someone read your message out of context and believe you are knocking NVidia.
Today Ed Greshko did spake thusly:
Scott van Looy wrote:
Just because this old card works for you at the moment doesn't mean it always will. NVidia are very good at supporting legacy hardware, other companies are less so, I was merely using them as an example as that's what we were talking about.
If/when it stops working I'm sure it will be due to the hardware crapping out.
Yes, NVidia is very good at supporting their legacy hardware...so it would have been appropriate to use a company that isn't, just to be fair and not have someone read your message out of context and believe you are knocking NVidia.
I _am_ knocking NVidia. They should open their damn drivers ;)
On Mon, 2007-10-29 at 15:29 +0800, Ed Greshko wrote:
Frank Cox wrote:
On Mon, 29 Oct 2007 00:36:47 -0500 Les Mikesell lesmikesell@gmail.com wrote:
There's a big difference in 'supporting' proprietary software and shooting the feet out from under your own users that depend on it.
If you depend on proprietary software, then Linux is not for you. Microsoft and/or Apple products are a much better fit for your needs.
Never did care that much for blanket statements.
If I depend on say "Oracle" and since Oracle is proprietary I should not run it on Linux?
RedHat's own internal support ticket system ran on Oracle. <huge grin> For awhile during the IPO days Oracle had at least 20 or so Engineers stationed at the old RTP location in Durham.
on 10/29/2007 10:44 PM, Les Mikesell wrote:
David Boles wrote:
on 10/29/2007 7:03 PM, Les Mikesell wrote:
David Boles wrote:
As for Fedora itself. Since you no longer use it why to you care anymore?
I have more than one machine, so I track fedora on a box that I can afford to have fail to boot on occasion to know what to expect from the next RHEL/Centos.
Ah! I see now. A 'bitch about it' box just for Fedora. ;-)
I look at it as having 3 chances to get the broken stuff fixed before I'm going to have to live with it in Centos... But, I'd give it much more of a workout if wasn't such a pain to keep updated under vmware since the interface to vmware tools changes all the time.
I thought that they, VMware, fixed that with 6.0x.
Version 6.0 of what? I don't see how they can fix it permanently without a sane interface on the Linux side to interoperate with.
Version 6.0 of VMware. Are you using an older version? There were some problems when 'they' (Linux folks) moved and renamed the kernel header files. It was fixed with a patch to the VMware config file. Either d/l and apply the patch or use your favorite text editor on the config. This was discussed to death, like this subject's thread, several times.
on 10/29/2007 9:03 PM, Tim wrote:
On Mon, 2007-10-29 at 10:20 -0500, Les Mikesell wrote:
One piece of software does not get to dictate what others you can run regardless of the unrealistic fanaticism involved.
Please explain that to Bill Gates...
Huh? Bill Gates has nothing to do with it. Microsoft wants software writers to test their product on Windows to make sure that it works and does not mess up the system before they can put the 'Windows ready' stuff on the box. What is wrong with that?
It that not the topic of many complaint threads here? I istalled 'this' and not my 'that' does not work?
Anders Karlsson wrote:
Thus, Les Mikesell at Mon Oct 29 23:45:49 2007 inscribed:
Craig White wrote:
I want something tested but not ancient. Neither disto provides that except for the first few months after an RHEL cut. But I'm usually happy with old kernel and server apps (except subversion and dovecot...). It's generally just the desktop stuff that changes fast enough to care about.
the problem is and will always be the interdependence among the various packages. While it may be convenient to look at things in a vacuum, they simply don't work that way - one package update requires updates on requisite packages which impacts something else down the line. The very thing that makes you rich, makes you poor.
That's a possibility, but rarely the case except perhaps for the Gnome/KDE environments themselves. Usually it is possible to build current packages on older RHEL versions, but then you have to maintain them yourself.
Let me see if I have understood this right. You look at subversion or dovecot in RHEL5, and you can determine by just looking at the version number that they are insufficient for your needs, even if they were suitable five months earlier, when RHEL5 was just released?
Not RHEL5, RHEL4 which is just barely middle-aged.
That, to me at least, would indicate that you have totally failed to understand the purpose of RHEL and the value that it provides to Enterprise customers.
There is much value in not having to update an OS and all the system libraries for several years and if the machine runs fine there should be no need to replace device drivers just to get a non-beta dovecot.
It also would indicate that you do not understand how Red Hat maintain each release of RHEL (although you are in good company there, as some customers don't understand that either to start with).
I do understand it - and I even like it for most server software because that has been feature-complete for years now. I was pointing out dovecot and subversion as rare exceptions on the server side just because of the progress they have made since the RHEL4 cut. On the desktop, though, everything is improving rapidly - but that still doesn't mean I want to have to replace working device drivers to get a new firefox.
Robert P. J. Day wrote:
> The difference is that closed source OS's rarely change their > driver interfaces, so it would be extremely unusual for > something that already works and I have put into production to > suddenly fail due to an update. I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
um ... no. the kernel *internal* interfaces may change, but the interface that is presented to the outside world is very stable.
http://lxr.linux.no/source/Documentation/stable_api_nonsense.txt
That's fine if you believe that the linux kernel will always include every device driver, filesystem, and feature you'll ever want. Don't ask me to join you in that leap of faith (or arrogance, or whatever it is that makes you think interoperability is unnecessary)...
what are you babbling about, les? all i did was clarify the distinction between the stability of the *internal* kernel interface and that of the *external* kernel interface. i said nothing whatsoever about driver support or interoperability.
Errr, what? How can you say that the internal kernel interface does not relate to the drivers that use them?
errr ... did you actually *read* the document for which i provided a URL, les? ***of course*** changing the internel kernel interfaces will undoubtedly affect the drivers which have been written to those interfaces.
Yes, I read it. That doesn't make it true, and my response applies.
and when those interfaces change, ***of course*** those drivers will have to be updated to conform to the new interfaces.
from the document above:
"Linux kernel development is continuous and at a rapid pace, never stopping to slow down. As such, the kernel developers find bugs in current interfaces, or figure out a better way to do things. If they do that, they then fix the current interfaces to work better. When they do so, function names may change, structures may grow or shrink, and function parameters may be reworked. If this happens, all of the instances of where this interface is used within the kernel are fixed up at the same time, ensuring that everything continues to work properly."
see how that works, les? when enough people in the kernel development community decide that an internal interface needs to change, everyone (theoretically) works together to make that transition as seamless and trouble-free as possible. and (note well) that all of that should be invisible to the outside world -- only driver developers need care. so what's your problem? what exactly bothers you about this process?
Exactly what I said before which apparently you didn't read. Neither experience nor blind faith lead me to believe that Linux will ever include all the drivers I might want. When Linus was just out of college he might have used the lack of experience to justify not wanting to freeze an interface because he might get it wrong. At this point I no longer believe that excuse. Interfaces should be treated as contracts among programmers that aren't whimsically ignored.
rday
p.s. also note, les, that if a given driver is in-tree, that driver's author doesn't even need to know about this, since it's the duty of those people making the change to *also* take care of all code that calls that interface. did you seriously think that someone might change an internal kernel interface and just leave all the broken calls to it hanging around in the kernel source tree?
Yes, using fedora I have had things break frequently. If you don't actually use an assortment of different hardware perhaps you haven't noticed. Even if I only used Linux on common hardware today, I would be concerned about depending on it and being unable to move to hardware with only vendor binary drivers in the future should that be the best hardware choice. And in my opinion this is the most significant reason that the majority of computers today don't run Linux.
Thus, Les Mikesell at Tue Oct 30 12:48:41 2007 inscribed:
Anders Karlsson wrote:
Thus, Les Mikesell at Mon Oct 29 23:45:49 2007 inscribed:
[snip]
That's a possibility, but rarely the case except perhaps for the Gnome/KDE environments themselves. Usually it is possible to build current packages on older RHEL versions, but then you have to maintain them yourself.
Let me see if I have understood this right. You look at subversion or dovecot in RHEL5, and you can determine by just looking at the version number that they are insufficient for your needs, even if they were suitable five months earlier, when RHEL5 was just released?
Not RHEL5, RHEL4 which is just barely middle-aged.
Even so, due to the fact that Red Hat backport features and fixes, you can not simply look at the version number to determine if the package is "out of date" or not.
That, to me at least, would indicate that you have totally failed to understand the purpose of RHEL and the value that it provides to Enterprise customers.
There is much value in not having to update an OS and all the system libraries for several years and if the machine runs fine there should be no need to replace device drivers just to get a non-beta dovecot.
Why would you need to replace device drivers if all you need to update is Dovecot? Dovecot does not have a dependency on the kernel. It is fully supported to take RHEL4 GA, install it, and then up2date dovecot to latest released through RHN. If you are happy with that system, and you do not care for any of the errata and enhancements released later, your system is still supported.
You'd be encouraged to update some packages if there are security vulnerabilities, but it's up to you if you want to or not. Support policy however is that if a later released version of a package contain a fix for the issue you report (should you report one) you'd be told to update that package, and its dependencies, to the version that contain the fix.
It also would indicate that you do not understand how Red Hat maintain each release of RHEL (although you are in good company there, as some customers don't understand that either to start with).
I do understand it - and I even like it for most server software because that has been feature-complete for years now. I was pointing out dovecot and subversion as rare exceptions on the server side just because of the progress they have made since the RHEL4 cut. On the desktop, though, everything is improving rapidly - but that still doesn't mean I want to have to replace working device drivers to get a new firefox.
And if you file RFE's (feature requests) you may get some of those new features and functionality backported as well. We're not psychic you know, you'd actually have to tell Red Hat that there is a feature you'd want to be considered for inclusion.
And again, firefox has to my knowledge not got a dependency on the kernel. You can take Firefox from the upcoming RHEL 4.6 and run that on an otherwise unmodified RHEL4GA. No problem.
Or maybe you mean you want to take Firefox from RHEL5, and run it on RHEL AS 2.1 as a supported setup?
Scott van Looy wrote:
Just because this old card works for you at the moment doesn't mean it always will. NVidia are very good at supporting legacy hardware, other companies are less so, I was merely using them as an example as that's what we were talking about.
If/when it stops working I'm sure it will be due to the hardware crapping out.
Yes, NVidia is very good at supporting their legacy hardware...so it would have been appropriate to use a company that isn't, just to be fair and not have someone read your message out of context and believe you are knocking NVidia.
I _am_ knocking NVidia. They should open their damn drivers ;)
Why, instead of you choosing an OS that cooperates with vendor efforts? First, they claim that they don't own what you want them to give you, and second, what basis do you have to tell them what to do even if that was a legal option for them?
On the other hand, if Linux were more cooperative with vendor drivers perhaps there would soon be a large enough base of users that would make it attractive to have a source-available driver and they would pay what it takes to obtain the right to do that.
on 10/30/2007 9:09 AM, Les Mikesell wrote:
Scott van Looy wrote:
I _am_ knocking NVidia. They should open their damn drivers ;)
Why, instead of you choosing an OS that cooperates with vendor efforts? First, they claim that they don't own what you want them to give you, and second, what basis do you have to tell them what to do even if that was a legal option for them?
On the other hand, if Linux were more cooperative with vendor drivers perhaps there would soon be a large enough base of users that would make it attractive to have a source-available driver and they would pay what it takes to obtain the right to do that.
So now you want Fedora to buy a license for a "source-available driver" and to include it in a Fedora release, and then *give* it to you? For free I gather? You are kidding here right? How much would you be willing to pay for a release like that if Fedora would do it?
BTW - When pigs fly.
If I had hardware that is as poorly supported as you claim your's is I would do one of two things. One. I would replace the hardware with something that works with Fedora. Two. I would search for a Linux distribution that works with my hardware.
One thing that i would *not* do. I would *not* do what you do on a regular basis. Come here with a 'poor me - you're picking on me' attitude. To date you are the *only* one that seems to have these problems consistently with Fedora. If I had problems like what you claim to have I would have left Fedora long, long ago.
Les Mikesell wrote:
Most of my closed source drivers are used with closed source OS's and are simply not an issue.
I assumed, since you are writing to a Fedora newsgroup/mailing list, that you were talking about using closed drivers under Fedora.
If you are talking about Windows drivers under Windows OS I don't really see the relevance.
Robert P. J. Day wrote:
I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
um ... no. the kernel *internal* interfaces may change, but the interface that is presented to the outside world is very stable.
http://lxr.linux.no/source/Documentation/stable_api_nonsense.txt
Did you actually read this article that you quoted? ------------------------------ You think you want a stable kernel interface, but you really do not, and you don't even know it. What you want is a stable running driver, and you get that only if your driver is in the main kernel tree. ------------------------------
As I understand it, this discussion is about drivers that are _not_ in the main kernel tree.
I quite often compile the orinoco_usb module, which is not in the kernel tree, and quite recent versions of this fail to compile with new kernels (eg 2.6.23 as compared with 2.6.20).
Today Les Mikesell did spake thusly:
Scott van Looy wrote:
Just because this old card works for you at the moment doesn't mean it always will. NVidia are very good at supporting legacy hardware, other companies are less so, I was merely using them as an example as that's what we were talking about.
If/when it stops working I'm sure it will be due to the hardware crapping out.
Yes, NVidia is very good at supporting their legacy hardware...so it would have been appropriate to use a company that isn't, just to be fair and not have someone read your message out of context and believe you are knocking NVidia.
I _am_ knocking NVidia. They should open their damn drivers ;)
Why, instead of you choosing an OS that cooperates with vendor efforts? First, they claim that they don't own what you want them to give you, and second, what basis do you have to tell them what to do even if that was a legal option for them?
FUD. Supposition.
On the other hand, if Linux were more cooperative with vendor drivers perhaps there would soon be a large enough base of users that would make it attractive to have a source-available driver and they would pay what it takes to obtain the right to do that.
I'll just stick to Intel/ATI, if it's all the same to you...
Timothy Murphy wrote:
Most of my closed source drivers are used with closed source OS's and are simply not an issue.
I assumed, since you are writing to a Fedora newsgroup/mailing list, that you were talking about using closed drivers under Fedora.
If you are talking about Windows drivers under Windows OS I don't really see the relevance.
The relevance is that the Linux faithful like to regurgitate the lines about how stable interfaces and binary drivers can't work when in fact they work just fine and the majority of the world runs on them taking advantage of the vendor's expertise and desire for a competitive advantage. Just a reality check...
Today Les Mikesell did spake thusly:
Timothy Murphy wrote:
Most of my closed source drivers are used with closed source OS's and are simply not an issue.
I assumed, since you are writing to a Fedora newsgroup/mailing list, that you were talking about using closed drivers under Fedora.
If you are talking about Windows drivers under Windows OS I don't really see the relevance.
The relevance is that the Linux faithful like to regurgitate the lines about how stable interfaces and binary drivers can't work when in fact they work just fine and the majority of the world runs on them taking advantage of the vendor's expertise and desire for a competitive advantage. Just a reality check...
They do? Most people here are trying to explain FOSS to you...
On Tue, 2007-10-30 at 05:45 -0400, David Boles wrote:
on 10/29/2007 10:44 PM, Les Mikesell wrote:
David Boles wrote:
on 10/29/2007 7:03 PM, Les Mikesell wrote:
David Boles wrote:
> As for Fedora itself. Since you no longer use it why to you care anymore? I have more than one machine, so I track fedora on a box that I can afford to have fail to boot on occasion to know what to expect from the next RHEL/Centos.
Ah! I see now. A 'bitch about it' box just for Fedora. ;-)
I look at it as having 3 chances to get the broken stuff fixed before I'm going to have to live with it in Centos... But, I'd give it much more of a workout if wasn't such a pain to keep updated under vmware since the interface to vmware tools changes all the time.
I thought that they, VMware, fixed that with 6.0x.
Version 6.0 of what? I don't see how they can fix it permanently without a sane interface on the Linux side to interoperate with.
Version 6.0 of VMware. Are you using an older version? There were some
VMware Workstation, in particular, is at version 6.0.2. Other VMware products have their own numbering schemes.
problems when 'they' (Linux folks) moved and renamed the kernel header files. It was fixed with a patch to the VMware config file. Either d/l and apply the patch or use your favorite text editor on the config. This was discussed to death, like this subject's thread, several times.
VMware modules currently build OK, but still need to be rebuilt at every kernel upgrade. If the module API changes again, then VMware users will all be back to using unofficial vmware-any-any patches to hack together something that works.
That's for the VM itself. VMware Tools (for the client machine) have the same issues: require rebuilding for kernel updates, patches when the module API changes.
On Tue, 30 Oct 2007, Timothy Murphy wrote:
Robert P. J. Day wrote:
I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
um ... no. the kernel *internal* interfaces may change, but the interface that is presented to the outside world is very stable.
http://lxr.linux.no/source/Documentation/stable_api_nonsense.txt
Did you actually read this article that you quoted?
You think you want a stable kernel interface, but you really do not, and you don't even know it. What you want is a stable running driver, and you get that only if your driver is in the main kernel tree.
As I understand it, this discussion is about drivers that are _not_ in the main kernel tree.
I quite often compile the orinoco_usb module, which is not in the kernel tree, and quite recent versions of this fail to compile with new kernels (eg 2.6.23 as compared with 2.6.20).
if you read my earlier post, i very explicitly mentioned that i was discussing in-tree drivers. if you insist on keeping your driver out of tree, then *of course* you'll run into trouble when internal kernel interfaces change. did you expect anything different?
rday
Robert P. J. Day wrote:
As I understand it, this discussion is about drivers that are _not_ in the main kernel tree.
I quite often compile the orinoco_usb module, which is not in the kernel tree, and quite recent versions of this fail to compile with new kernels (eg 2.6.23 as compared with 2.6.20).
if you read my earlier post, i very explicitly mentioned that i was discussing in-tree drivers.
A curious thing, given that the discussion is about the things not supported in-tree - or badly supported.
if you insist on keeping your driver out of tree,
It's not 'my' driver in or out of the tree. I just want to be able to use the hardware and additional software of my choice.
then *of course* you'll run into trouble when internal kernel interfaces change. did you expect anything different?
Yes, I did expect something different, given Linus's initial acceptance of proprietary kernel modules. I thought his initial refusal to freeze an interface was simply based on inexperience and that eventually he would be able to design something permanent. Obviously I was mistaken or mislead.
Scott van Looy wrote:
Today Les Mikesell did spake thusly:
Timothy Murphy wrote:
Most of my closed source drivers are used with closed source OS's and are simply not an issue.
I assumed, since you are writing to a Fedora newsgroup/mailing list, that you were talking about using closed drivers under Fedora.
If you are talking about Windows drivers under Windows OS I don't really see the relevance.
The relevance is that the Linux faithful like to regurgitate the lines about how stable interfaces and binary drivers can't work when in fact they work just fine and the majority of the world runs on them taking advantage of the vendor's expertise and desire for a competitive advantage. Just a reality check...
They do?
http://www.w3schools.com/browsers/browsers_os.asp puts Linux at 3%. Somewhere I thought I saw that current sales were around 8% Mac compared to the overall 3.8% though.
Most people here are trying to explain FOSS to you...
The issue has nothing to do with open source, which in general co-exists nicely with proprietary software.
On Tue, 2007-10-30 at 11:32 -0500, Les Mikesell wrote:
Scott van Looy wrote:
Today Les Mikesell did spake thusly:
Timothy Murphy wrote:
Most of my closed source drivers are used with closed source OS's and are simply not an issue.
I assumed, since you are writing to a Fedora newsgroup/mailing list, that you were talking about using closed drivers under Fedora.
If you are talking about Windows drivers under Windows OS I don't really see the relevance.
The relevance is that the Linux faithful like to regurgitate the lines about how stable interfaces and binary drivers can't work when in fact they work just fine and the majority of the world runs on them taking advantage of the vendor's expertise and desire for a competitive advantage. Just a reality check...
They do?
http://www.w3schools.com/browsers/browsers_os.asp puts Linux at 3%. Somewhere I thought I saw that current sales were around 8% Mac compared to the overall 3.8% though.
---- those numbers just don't look right. I refuse to believe that Win98 is and has been under 1% and Vista is so small.
As for Macintosh sales being 8%, I haven't seen any such report...that would be a significant increase.
Craig
Craig White wrote:
The relevance is that the Linux faithful like to regurgitate the lines about how stable interfaces and binary drivers can't work when in fact they work just fine and the majority of the world runs on them taking advantage of the vendor's expertise and desire for a competitive advantage. Just a reality check...
They do?
http://www.w3schools.com/browsers/browsers_os.asp puts Linux at 3%. Somewhere I thought I saw that current sales were around 8% Mac compared to the overall 3.8% though.
those numbers just don't look right. I refuse to believe that Win98 is and has been under 1% and Vista is so small.
Vista has not been well accepted at the enterprise level. Something about changing driver interfaces, perhaps... Microsoft has been forced to extend their support for XP and Dell continues to offer it.
As for Macintosh sales being 8%, I haven't seen any such report...that would be a significant increase.
It is significant. They are up 34% this quarter from a year ago with the PC market only growing at about 15% http://www.news.com/8301-13579_3-9801997-37.html And with Leopard they are officially Posix compliant and UNIX 03 registered. And they have dtrace.
On Monday 29 October 2007, Anders Karlsson wrote:
Thus, Les Mikesell at Mon Oct 29 22:07:52 2007 inscribed:
I think you misunderstood. I said closed source - as opposed to Linux - doesn't change driver interfaces often. With Linux the kernel changes continuously but it is up to the distribution what is shipped. RHEL maintains something stable. Fedora doesn't.
The support and the ABI/API stability is part of what you pay for with your subscription with RHEL. You pay nothing for Fedora. Stop trying to turn Fedora in to RHEL or CentOS, they have different purposes and goals.
What Les would like is a Fedora on a more stable (as in driver/module INTERNAL API).
The discussion always seems to morph here, for some reason. CentOS/RHEL provide API stability (in general) across the entire OS, down to the version numbers. Fedora doesn't provide this, for any package, including the kernel. The kernel's module/driver INTERNAL interface changes regularly, sometimes for no good reason it seems.
This is an upstream kernel development problem, not a Fedora one.
I too get quite aggravated by this. No, just because I choose to run a few proprietary modules in the kernel (in my case, it's all vmware stuff) does not mean at all that I shouldn't run Linux, or even Fedora, for that matter. Perhaps I'm not even running VMware for the reasons you think I am?
FOSS and proprietary software are not and will not be mutually exclusive, despite some folks misplaced idealism.
Robert P. J. Day wrote:
On Tue, 30 Oct 2007, Les Mikesell wrote:
Robert P. J. Day wrote:
On Tue, 30 Oct 2007, Les Mikesell wrote:
Robert P. J. Day wrote:
> The difference is that closed source OS's rarely change their > driver interfaces, so it would be extremely unusual for > something that already works and I have put into production to > suddenly fail due to an update. > I find this an astonishing assertion. Surely the Linux kernel interface changes reasonably often?
um ... no. the kernel *internal* interfaces may change, but the interface that is presented to the outside world is very stable.
http://lxr.linux.no/source/Documentation/stable_api_nonsense.txt
That's fine if you believe that the linux kernel will always include every device driver, filesystem, and feature you'll ever want. Don't ask me to join you in that leap of faith (or arrogance, or whatever it is that makes you think interoperability is unnecessary)...
what are you babbling about, les? all i did was clarify the distinction between the stability of the *internal* kernel interface and that of the *external* kernel interface. i said nothing whatsoever about driver support or interoperability.
Errr, what? How can you say that the internal kernel interface does not relate to the drivers that use them?
errr ... did you actually *read* the document for which i provided a URL, les? ***of course*** changing the internel kernel interfaces will undoubtedly affect the drivers which have been written to those interfaces. and when those interfaces change, ***of course*** those drivers will have to be updated to conform to the new interfaces.
from the document above:
"Linux kernel development is continuous and at a rapid pace, never stopping to slow down. As such, the kernel developers find bugs in current interfaces, or figure out a better way to do things. If they do that, they then fix the current interfaces to work better. When they do so, function names may change, structures may grow or shrink, and function parameters may be reworked. If this happens, all of the instances of where this interface is used within the kernel are fixed up at the same time, ensuring that everything continues to work properly."
see how that works, les? when enough people in the kernel development community decide that an internal interface needs to change, everyone (theoretically) works together to make that transition as seamless and trouble-free as possible. and (note well) that all of that should be invisible to the outside world -- only driver developers need care. so what's your problem? what exactly bothers you about this process?
rday
p.s. also note, les, that if a given driver is in-tree, that driver's author doesn't even need to know about this, since it's the duty of those people making the change to *also* take care of all code that calls that interface. did you seriously think that someone might change an internal kernel interface and just leave all the broken calls to it hanging around in the kernel source tree?
A while ago a very smart man told me that you NEVER change a kernel unless you need something the new kernel provides. If the one I have now that I got 1 day ago really do ME more good? NO.
Thus, Lamar Owen at Tue Oct 30 20:20:08 2007 inscribed: [snip]
What Les would like is a Fedora on a more stable (as in driver/module INTERNAL API).
The discussion always seems to morph here, for some reason. CentOS/RHEL provide API stability (in general) across the entire OS, down to the version numbers. Fedora doesn't provide this, for any package, including the kernel. The kernel's module/driver INTERNAL interface changes regularly, sometimes for no good reason it seems.
This is an upstream kernel development problem, not a Fedora one.
Agreed, which is why I earlier in the thread suggested that the discussion would be better off on the LKML, where the kernel people that makes those decisions could explain it to the people arguing the point about this.
I too get quite aggravated by this. No, just because I choose to run a few proprietary modules in the kernel (in my case, it's all vmware stuff) does not mean at all that I shouldn't run Linux, or even Fedora, for that matter. Perhaps I'm not even running VMware for the reasons you think I am?
Your reason for running VMware is not something that I care much about. When you require support for a kernel crash, I'd expect you to be asked to remove the proprietary, closed sourced, drivers/modules and reproduce before re-reporting a problem. This is just common sense really.
FOSS and proprietary software are not and will not be mutually exclusive, despite some folks misplaced idealism.
There is a distinction to be made. When you talk about proprietary, closed source, drivers - then people should be aware that they not just may, they *will*, get strong reactions.
I hear what you are saying though.
Anders Karlsson wrote:
What Les would like is a Fedora on a more stable (as in driver/module INTERNAL API).
The discussion always seems to morph here, for some reason. CentOS/RHEL provide API stability (in general) across the entire OS, down to the version numbers. Fedora doesn't provide this, for any package, including the kernel. The kernel's module/driver INTERNAL interface changes regularly, sometimes for no good reason it seems.
This is an upstream kernel development problem, not a Fedora one.
Agreed, which is why I earlier in the thread suggested that the discussion would be better off on the LKML, where the kernel people that makes those decisions could explain it to the people arguing the point about this.
No, the kernel people are and should be free to write total garbage. The distributions shouldn't ship that to unsuspecting users, though.
FOSS and proprietary software are not and will not be mutually exclusive, despite some folks misplaced idealism.
There is a distinction to be made. When you talk about proprietary, closed source, drivers - then people should be aware that they not just may, they *will*, get strong reactions.
I happen to love my 1st gen tivo. I suppose you'll try to tell me its a bad thing that they exist.
Thus, Les Mikesell at Wed Oct 31 01:13:24 2007 inscribed:
Anders Karlsson wrote:
[snip]
Agreed, which is why I earlier in the thread suggested that the discussion would be better off on the LKML, where the kernel people that makes those decisions could explain it to the people arguing the point about this.
No, the kernel people are and should be free to write total garbage. The distributions shouldn't ship that to unsuspecting users, though.
You are more than welcome to contribute to Fedora or the kernel. If coding is not for you, then perhaps QA or packaging is more to taste? The effort you'd put in would potentially save some unsuspecting users from the garbage that you are talking about.
What I do not yet understand, and please do enlighten me, is why you are still here considering the appalling experiences you have had with Linux, Fedora etc., over what seems to be rather extensive periods of time. I admire the staying-power, as most people would have given up years ago, given the totallly disastrous experiences you seem to have had.
I mean, surely you'd be contributing to and working closely with the Hurd project as the Linux kernel is such a pile of pap?
There is a distinction to be made. When you talk about proprietary, closed source, drivers - then people should be aware that they not just may, they *will*, get strong reactions.
I happen to love my 1st gen tivo. I suppose you'll try to tell me its a bad thing that they exist.
Not a bad thing that they exist, no. As long as they adhere to the various licenses, such as the GPL, for the components that they have incorporated into their product.
There was the empeg car-stereo project as well, now that was seriously awesome, *and* they stuck to the licenses too. How about that?
Les Mikesell wrote:
Anders Karlsson wrote:
[...]
Agreed, which is why I earlier in the thread suggested that the discussion would be better off on the LKML, where the kernel people that makes those decisions could explain it to the people arguing the point about this.
No, the kernel people are and should be free to write total garbage. The distributions shouldn't ship that to unsuspecting users, though.
I'm sure you have some idea of how much effort it takes for RHEL to maintain a stable interface for the life of a release. That effort wouldn't happen without a lot of hours and dollars. Those resources are not available in the needed quantities in Fedora.
It's also clearly a goal of the Fedora project to follow upstream, including the kernel. I was glad when my SD card reader began working during the FC5 cycle with a kernel update. Many other folks are happy with the improvements brought in by new kernels.
If you don't want to update your kernel, add an exclude to the yum.conf.
If you want to volunteer to help maintain a stable kernel interface in Fedora, please do so on the kernel list.
But seriously, what gain is there to posting about your dislike of the kernel's interface policy on this list, repeatedly? Do you think it's likely to bring about the changes you are seeking?
On Tue, 2007-10-30 at 13:22 -0500, Les Mikesell wrote:
Craig White wrote:
The relevance is that the Linux faithful like to regurgitate the lines about how stable interfaces and binary drivers can't work when in fact they work just fine and the majority of the world runs on them taking advantage of the vendor's expertise and desire for a competitive advantage. Just a reality check...
They do?
http://www.w3schools.com/browsers/browsers_os.asp puts Linux at 3%. Somewhere I thought I saw that current sales were around 8% Mac compared to the overall 3.8% though.
those numbers just don't look right. I refuse to believe that Win98 is and has been under 1% and Vista is so small.
Vista has not been well accepted at the enterprise level. Something about changing driver interfaces, perhaps... Microsoft has been forced to extend their support for XP and Dell continues to offer it.
As for Macintosh sales being 8%, I haven't seen any such report...that would be a significant increase.
It is significant. They are up 34% this quarter from a year ago with the PC market only growing at about 15% http://www.news.com/8301-13579_3-9801997-37.html And with Leopard they are officially Posix compliant and UNIX 03 registered. And they have dtrace.
---- 34% increase in Apple bluster. Article was about iPhones/iPods
Nothing there suggested any Apple percentage of total desktop sales
I wonder how much device driver breakage there is with Leopard. ;-) In fact, I have not considered any Apple OSX to be what I would call a stable environment, I actually refer to it as permanent beta.
Not to mention the Apple Tax so you can pay for an OS when you already have an OS.
Finally, I got my mailing from Tidbits and it seems that all is definitely not rosy with Leopard
Craig
Craig White wrote:
On Tue, 2007-10-30 at 13:22 -0500, Les Mikesell wrote:
Craig White wrote:
The relevance is that the Linux faithful like to regurgitate the lines about how stable interfaces and binary drivers can't work when in fact they work just fine and the majority of the world runs on them taking advantage of the vendor's expertise and desire for a competitive advantage. Just a reality check...
They do?
http://www.w3schools.com/browsers/browsers_os.asp puts Linux at 3%. Somewhere I thought I saw that current sales were around 8% Mac compared to the overall 3.8% though.
those numbers just don't look right. I refuse to believe that Win98 is and has been under 1% and Vista is so small.
Vista has not been well accepted at the enterprise level. Something about changing driver interfaces, perhaps... Microsoft has been forced to extend their support for XP and Dell continues to offer it.
As for Macintosh sales being 8%, I haven't seen any such report...that would be a significant increase.
It is significant. They are up 34% this quarter from a year ago with the PC market only growing at about 15% http://www.news.com/8301-13579_3-9801997-37.html And with Leopard they are officially Posix compliant and UNIX 03 registered. And they have dtrace.
34% increase in Apple bluster. Article was about iPhones/iPods
Quote: "Mac shipments were up 34 percent compared to last year"
Nothing there suggested any Apple percentage of total desktop sales
Quote: "It's also clear that Apple is gaining share on the rest of the PC industry. Last week IDC and Gartner had the worldwide PC market growing at around 15 percent, while Mac shipments are growing more than twice as fast. Apple sold 2.1 million Macs during the quarter, a company record and 400,000 units better than its previous best"
I wonder how much device driver breakage there is with Leopard. ;-)
I haven't heard of any so far. And keep in mind that the same thing runs on both PPC and intel processors, unlike anything else.
In fact, I have not considered any Apple OSX to be what I would call a stable environment, I actually refer to it as permanent beta.
I haven't had anything break in several years (longer if you count a g3 powermac at work). And I'm good at breaking things.
Not to mention the Apple Tax so you can pay for an OS when you already have an OS.
Can't argue with that, but the family pack at $199 for 5 licenses isn't that bad and a reasonable tradeoff for things that "just work". How much is it for one RHEL these days? Plus it is worth something to be able to legally play dvd's and mp3s.
Finally, I got my mailing from Tidbits and it seems that all is definitely not rosy with Leopard
I haven't updated yet but so far I haven't heard of any showstoppers. zfs didn't make it in though, which is a little disappointing.
On Tue, 2007-10-30 at 22:02 -0500, Les Mikesell wrote:
Craig White wrote:
On Tue, 2007-10-30 at 13:22 -0500, Les Mikesell wrote:
Craig White wrote:
> The relevance is that the Linux faithful like to regurgitate the lines > about how stable interfaces and binary drivers can't work when in fact > they work just fine and the majority of the world runs on them taking > advantage of the vendor's expertise and desire for a competitive > advantage. Just a reality check... They do?
http://www.w3schools.com/browsers/browsers_os.asp puts Linux at 3%. Somewhere I thought I saw that current sales were around 8% Mac compared to the overall 3.8% though.
those numbers just don't look right. I refuse to believe that Win98 is and has been under 1% and Vista is so small.
Vista has not been well accepted at the enterprise level. Something about changing driver interfaces, perhaps... Microsoft has been forced to extend their support for XP and Dell continues to offer it.
As for Macintosh sales being 8%, I haven't seen any such report...that would be a significant increase.
It is significant. They are up 34% this quarter from a year ago with the PC market only growing at about 15% http://www.news.com/8301-13579_3-9801997-37.html And with Leopard they are officially Posix compliant and UNIX 03 registered. And they have dtrace.
34% increase in Apple bluster. Article was about iPhones/iPods
Quote: "Mac shipments were up 34 percent compared to last year"
---- that's still an unsubstantiated Apple claim which has no relation to market percentage itself. ----
Nothing there suggested any Apple percentage of total desktop sales
Quote: "It's also clear that Apple is gaining share on the rest of the PC industry. Last week IDC and Gartner had the worldwide PC market growing at around 15 percent, while Mac shipments are growing more than twice as fast. Apple sold 2.1 million Macs during the quarter, a company record and 400,000 units better than its previous best"
Not to mention the Apple Tax so you can pay for an OS when you already have an OS.
Can't argue with that, but the family pack at $199 for 5 licenses isn't that bad and a reasonable tradeoff for things that "just work". How much is it for one RHEL these days? Plus it is worth something to be able to legally play dvd's and mp3s.
---- You can't use the 5 pack in businesses, they are specifically excluded. In fact, that is discriminatory pricing which really doesn't make much sense at all.
I don't watch dvd's on my computer and used my Windows iTunes to rip my CD's into iTunes (legally). I used Linux to rip my DVD's into m4v/mp4 format...does that make me a lawbreaker? ----
Finally, I got my mailing from Tidbits and it seems that all is definitely not rosy with Leopard
I haven't updated yet but so far I haven't heard of any showstoppers. zfs didn't make it in though, which is a little disappointing.
---- showstoppers is always relative...I have to support the suckers.
http://db.tidbits.com/issue/902 This is from people who love Macintosh
I used to be vp of Apple user group...I am so disenchanted with Apple. The way I figure things, there is little difference between Apple and Microsoft with the sole exception being that Microsoft has enjoyed tremendous success selling desktop and server OS's.
- They couldn't care less about businesses.
- Have to pay $100 annual extortion fee to get systems serviced in under 2 weeks.
- Their hardware is terrific but excessively priced (all Apple mice excepted). In fact, all of the Apple users tossed out their Apple 'mighty mouse' in favor of cheap, Kensington optical mouse.
- Their user interface is entirely confused, ever changing, inconsistent, disadvantaged.
- Single button mouse, finder that continually morphs.
- Sherlock has continually sucked. Dock sucks. Top Menu bar sucks. Worst thing about Macintosh...users have to be told where their files are saved - incredible.
- They've done nothing to further development of OpenOffice.org software.
- LDAP automount of home directories requires - of all things...an unauthenticated initial connection to the Linux server via AppleTalk (yeah, I compiled AppleTalk kernel module), and installing updates (10.4.x) shuts AppleTalk off and users can't mount their home directories until I log in as Administrator, turn AppleTalk back on and then log off again...perpetual beta. Probably not much hope that they've changed this on Leopard but I am not encouraging them to update at this point either.
Hardly surprising that a whole bunch of updates have already come out for Leopard...perpetual beta
The last 4 new Mac's that my office brought in took 3.5 hours to download and install updates onto shipped OS (time included 2 restarts).
Craig
Craig White wrote:
I don't watch dvd's on my computer and used my Windows iTunes to rip my CD's into iTunes (legally). I used Linux to rip my DVD's into m4v/mp4 format...does that make me a lawbreaker?
Yes, at least if you are in the US. There's probably a case to be made for personal use of the DVD content, but not for using unlicensed software that uses techniques covered by patents to do it.
I used to be vp of Apple user group...I am so disenchanted with Apple.
I don't get it. The original Macs were useless too-small closed boxes with a single tasking OS. Now they are some of the fastest things around, only slightly overpriced, run a full unix, and can run windows, linux or solaris in virtual machines under parallels or vmware. I don't see a down side there.
The way I figure things, there is little difference between Apple and Microsoft with the sole exception being that Microsoft has enjoyed tremendous success selling desktop and server OS's.
And Apple has a real unix underneath.
- They couldn't care less about businesses.
But you can compile and run all of your unix code.
- Have to pay $100 annual extortion fee to get systems serviced in under
2 weeks.
Keeping spares works for PC's, it would probably work for Macs too.
- Their hardware is terrific but excessively priced (all Apple mice
excepted).
But since all prices have come down, it doesn't matter so much now.
In fact, all of the Apple users tossed out their Apple 'mighty mouse' in favor of cheap, Kensington optical mouse.
I've sort-of gotten used to the mighty mouse.
- Their user interface is entirely confused, ever changing,
inconsistent, disadvantaged.
You can always run X base software if you prefer. I usually run firefox and thunderbird so the location of the close/minimize buttons is the biggest change when OS hopping.
- Single button mouse, finder that continually morphs.
The mighty mouse can do left/right clicks if you change the setting.
- Sherlock has continually sucked.
What's a Sherlock?
Dock sucks. Top Menu bar sucks.
So so... It's all pretty arbitrary.
Worst thing about Macintosh...users have to be told where their files are saved - incredible.
You have it backwards. Having to ever know where the files are is the problem - if that's the case for any of the included software. For example, if you want to copy some files from itunes or iphoto you wouldn't go find the files themselves, you'd use the application features to get the songs or photos listed where you could select them and drag the titles or pictures to a folder/flash drive, etc. If you are going to use a GUI at all you should go all the way and not care about locations or filenames.
- They've done nothing to further development of OpenOffice.org
software.
Why should they, any more than Dell would? In fact they have their own 'pages' which would be a competitor.
- LDAP automount of home directories requires - of all things...an
unauthenticated initial connection to the Linux server via AppleTalk (yeah, I compiled AppleTalk kernel module), and installing updates (10.4.x) shuts AppleTalk off and users can't mount their home directories until I log in as Administrator, turn AppleTalk back on and then log off again...perpetual beta. Probably not much hope that they've changed this on Leopard but I am not encouraging them to update at this point either.
Is this a Posix requirement or something in the UNIX spec? If so it should be in Leopard. If its not in a standards spec, I don't understand the complaint.
Hardly surprising that a whole bunch of updates have already come out for Leopard...perpetual beta
What?
The last 4 new Mac's that my office brought in took 3.5 hours to download and install updates onto shipped OS (time included 2 restarts).
In the time span from that OS release you would have had to download and install at least 2 versions of fedora, each of which would have had about 10x the updates to download. The difference is that some of the fedora updates wouldn't have booted on some hardware.
On Tue, 2007-10-30 at 14:27 -0600, Karl Larsen wrote:
A while ago a very smart man told me that you NEVER change a
kernel unless you need something the new kernel provides. If the one I have now that I got 1 day ago really do ME more good? NO.
Though, you may find that you want some other updated package, and that may require the current kernel. It's not always so easy to sort out, you can't just decide not to update kernels because you don't want to, you have to think the whole thing through.
On Mon, 2007-10-29 at 01:38 -0600, Frank Cox wrote:
On Mon, 29 Oct 2007 15:29:46 +0800 Ed Greshko Ed.Greshko@greshko.com wrote:
If you depend on proprietary software, then Linux is not for you. Microsoft and/or Apple products are a much better fit for your needs.
Never did care that much for blanket statements.
If I depend on say "Oracle" and since Oracle is proprietary I should not run it on Linux?
You are taking a general statement and applying it to a very specific circumstance.
In the particular case cited, Linux may indeed be the answer you seek. However, in general terms, the solutions that folks like Les are looking for can be most easily found in places other than Linux. Microsoft and Apple are the two largest examples of such places.
Frank, since when have you become a mind reader? I think you have some notion that you can see clearly into his brain and can speak authoritatively on what he thinks.
I know Les very well. He's got some coding chops I'd die for. We both ran into the same wall with the same application and the same identical nVidia 5200 video cards. The nvidia rpms supplied by our 3rd party sites are lacking SOMETHING with regards to openGL. I beat myself up for months, trying to get the app to run correctly. So, Les volunteered out of the blue to figure out what was wrong, from the very goodness of his heart. I love him as a Brother for it. He couldn't get it to work either. We both kept running into openGL problems and the Windows / Mac users had reported no problems at all. Emails flew back and forth between us trying to trouble shoot the damn thing.
I don't know about you, but I bristle at the notion that they could make happen so easily what had us completely stumped. So, I raged and swore, determined that Linux, as a community, was not to be relegated to the cheap seats as lusers. So, I finally beat on Lonnie and he told me to install the nVidia package from them.
Here's the saddest part of the ordeal. Reading all of this "my-way or the highway" crap put me off from considering using something as EVIL as a proprietary driver. Bad! Bad! No! No! Well, I finally tried it. It worked. I told Les, he tried it, it worked. End of story. I lost a GD pant-load of valuable learning and devel time ...several **months** at least. Les lost quite a bit of his donated personal time too. THAT is where he's coming from, and so do I.
I'm trying to develop a 3D educational environment aimed at reducing the recidivism rate of prisoners coming out of prison back to the "Real World". Les has helped out immeasurably. "Polite Society" is not kind to guys and gals getting out. Inside the razor wire when you disrespect someone, predictably you get punched in the face ...*real quick*. Take someone that has lived like that for 5, 10 or 20 years, someone better de-fuse him before he leaves. OR someone on this side of the wire gets hurt to become the next new victim ...to the tune of 70 new victims for every 100 released. You think I give one shit about some evil proprietary driver? Should I when we have 1/4 million inmates being released per year over the next 10 years and I really want this effort to work in an otherwise Open-Source Fashion and keep Gates, Jobs and their ilk off of it?
I see the day coming when Linux will have something like CUPS for video and everything will be peaches and cream. Like in the old days, we had the choice of LPR (great for Daisy Wheels and line printers) or the proprietary version of Adobe, if you wanted your nifty laser printer to work at all. Everyone hated that too, back when. Same with OSS, it filled a niche allowing some people (like myself) to make an odd-ball (PAS16) sound card work when the kernel didn't seem to make it happen correctly or at all. I paid my ten bucks to Hannu and was as happy as a clam as the solution was quite cheap and worked beautifully. Back then no one had a problem relaying this information to another in need of it, either. Even at Red Hat. Later on, the problem became fixed and continues to improve all the time. I like that and appreciate the work by the people that make it happen.
Alan, remember when the only Windows Machine at Red Hat was the one that made the CD dupes, printed the labels and stuck them on automatically?? It was a collective groan (and well kept secret) for sure, but we lived with it, proprietary and all. It got the job done that needed to be done, and the only alternative was to pay someone 40 grand a year or more to sit on a stool in the old "Red Hat Ready" hardware evaluation area and do it all by hand ...one at a time. Damn skippy, the proprietary solution at the time wasn't so EVIL that it could NOT be tolerated, was it? <wink wink> You know I'm right and I just told the world to make the point.
The issue is not that Les or I insist that nVidia be installed on the Dvd (which he has never written) but that when the need arises that folks don't feel all intimidated from giving useful information that may be needed, even if a couple of K-bytes get used on the list to convey the information without mega-bytes of retaliation. I do know that running software openGL is slower than hardware openGL. That is a "truth" ... which is the "Right Thing(tm)". Becoming at least par with the average Windows or Mac user is a good thing, too. Now more people are less afraid to use Linux on the Croquet list after Les and I reported the solution to their developers and their list members.
We're discipling Linux, not condemning it. But if my personal need is great enough, then I'll alter my FC install, taint the kernel and be responsible for any problems that arise from it ...if only to make my computer do what I need it to do. Ric
On Wed, 31 Oct 2007 02:59:00 -0400 Ric Moore wayward4now@gmail.com wrote:
Frank, since when have you become a mind reader? I think you have some notion that you can see clearly into his brain and can speak authoritatively on what he thinks.
I have been carefully reading his comments and replying to what he says. If he is capable of expressing his thoughts in a clear manner, then that's what I've been responding to. Mind reading is not required -- simply reading, period.
Here's the saddest part of the ordeal.
I'd agree with that. Pretty sad, indeed.
I lost a GD pant-load of valuable learning and devel time ...several **months** at least. Les lost quite a bit of his donated personal time too. THAT is where he's coming from, and so do I.
Sounds like you have a lot to thank Nvidia for here. Months of time lost to a poorly implemented video driver as a starting point.
Were I you, I'd not be saying "Please sir, may I have another" after going through that experience. I'd be moving on to something that works properly out-of-the-box. That means Intel at the moment and ATI shortly (apparently).
On Mon, 2007-10-29 at 10:19 -0700, Craig White wrote:
On Mon, 2007-10-29 at 03:21 -0400, Ric Moore wrote:
On Sun, 2007-10-28 at 20:05 -0700, Craig White wrote:
On Sun, 2007-10-28 at 22:37 -0400, David Boles wrote:
What is your latest disaster this time anyway? I missed that part of this thread.
Les' mission is to save us from the evils of GPL license.
Not so, no where did Les say that, Craig. I know Les ...he has a bunch of coding chops under his belt and he's a helluva decent human being. The point is that it needn't be anathema if the only way to get a device working correctly is to use a driver from the manufacturer. Sure, pang on them until the cows come home to open up their code. I believe in that. One day everyone will see * * The Light * *. Guys like Alan Cox could tell them a thing or three or ten about proper interfacing to the kernel.
Meanwhile, getting stuff to work is a consideration for Joe Lunch de Bucket, Les, me and a bunch of others. And, we deal with what we got as we can. :) Ric
evidently you don't know Les all that well or simply haven't tracked his feelings about GPL
Read the book I just wrote. I know Les pretty darn well, for never having met him in person. I understand where he is coming from and it takes just a perception shift to understand Les. He's an old school ex-navy coder that takes a very personal interest in computing systems. He's not some script kiddie or Joe Lunchbucket whining about the latest space game not working.
His position is the same as mine that if the GPL stops you **utterly in your tracks** from doing useful stuff, then the old-school "Right Thing" principle kicks in. You look for a solution, any solution to get the job done. Something is rotten in Denmark when the goal is being righteous as opposed to being Right. I used to think that the old Debian crowd were the only zealots in Linux Land. If someone in the know on this list had told me that there are proprietary bits missing and that to solve the problem would be to use the nVidia supplied driver, I would be several months ahead of where I am now. I'm royally jerked that someone knew this and didn't tell me on the list out of Fear, Uncertainty and Doubt. Period. The little bit of pay that I get comes from donations to our Non-Profit, from some great people and almost two months worth was wasted while I was sweating bullets. See why I'm pissed? I thank GOD that Les stepped up to help me regain lost ground. He really knows his stuff. Far more than you would guess. I thank Lonnie for cluing me in as well. He was right, as my problem vanished immediately.
OTOH, I will email nVidia (while I know that Lonnie is in lurking mode here) and ask that they at least Open Source the the older hardware code where the added benefit is that they don't have to provide nor maintain it anymore. They can do the same for the Winders crowd. What say you? I think that would be a dandy idea with the added benefit that NVidia just upped the ante on their competitors. Any free press is GOOD PRESS! Right Lonnie? I know you're there! Make this thing happen, if you can. That will also be the "Right Thing". Ric
Somebody in the thread at some point said:
His position is the same as mine that if the GPL stops you **utterly in your tracks** from doing useful stuff, then the old-school "Right Thing" principle kicks in. You look for a solution, any solution to get the job done. Something is rotten in Denmark when the goal is being righteous as
The GPL doesn't stop you combining stuff, it stops you distributing some combinations. That's a bit different. Proprietary licensed stuff is far worse, you can't redistribute it AT ALL or you get chucked in jail. This and his unwillingness to converge on any agreement different from his starting position is my problem with Les' complaints, despite I have seen him say a lot of smart things otherwise.
Sometimes you have to use proprietary stuff to do what you need to do at all -- but it wasn't that long ago that 100% proprietary was the only choice. Now for a lot of us it is the unusual case at the edge on our PCs and even outside of PCs FOSS solutions are creeping into the other devices around us too.
But choosing to use a proprietary solution as the enduser is different from changing the basis of a Free distro to include it. Whereas we individually might have to compromise to go on, the ideal shouldn't be compromised (he said, trying not to think about firmware exceptions).
-Andy
On Wed, 2007-10-31 at 03:42 -0400, Ric Moore wrote:
On Mon, 2007-10-29 at 10:19 -0700, Craig White wrote:
On Mon, 2007-10-29 at 03:21 -0400, Ric Moore wrote:
On Sun, 2007-10-28 at 20:05 -0700, Craig White wrote:
On Sun, 2007-10-28 at 22:37 -0400, David Boles wrote:
What is your latest disaster this time anyway? I missed that part of this thread.
Les' mission is to save us from the evils of GPL license.
Not so, no where did Les say that, Craig. I know Les ...he has a bunch of coding chops under his belt and he's a helluva decent human being. The point is that it needn't be anathema if the only way to get a device working correctly is to use a driver from the manufacturer. Sure, pang on them until the cows come home to open up their code. I believe in that. One day everyone will see * * The Light * *. Guys like Alan Cox could tell them a thing or three or ten about proper interfacing to the kernel.
Meanwhile, getting stuff to work is a consideration for Joe Lunch de Bucket, Les, me and a bunch of others. And, we deal with what we got as we can. :) Ric
evidently you don't know Les all that well or simply haven't tracked his feelings about GPL
Read the book I just wrote. I know Les pretty darn well, for never having met him in person. I understand where he is coming from and it takes just a perception shift to understand Les. He's an old school ex-navy coder that takes a very personal interest in computing systems. He's not some script kiddie or Joe Lunchbucket whining about the latest space game not working.
His position is the same as mine that if the GPL stops you **utterly in your tracks** from doing useful stuff, then the old-school "Right Thing" principle kicks in. You look for a solution, any solution to get the job done. Something is rotten in Denmark when the goal is being righteous as opposed to being Right. I used to think that the old Debian crowd were the only zealots in Linux Land. If someone in the know on this list had told me that there are proprietary bits missing and that to solve the problem would be to use the nVidia supplied driver, I would be several months ahead of where I am now. I'm royally jerked that someone knew this and didn't tell me on the list out of Fear, Uncertainty and Doubt. Period. The little bit of pay that I get comes from donations to our Non-Profit, from some great people and almost two months worth was wasted while I was sweating bullets. See why I'm pissed? I thank GOD that Les stepped up to help me regain lost ground. He really knows his stuff. Far more than you would guess. I thank Lonnie for cluing me in as well. He was right, as my problem vanished immediately.
OTOH, I will email nVidia (while I know that Lonnie is in lurking mode here) and ask that they at least Open Source the the older hardware code where the added benefit is that they don't have to provide nor maintain it anymore. They can do the same for the Winders crowd. What say you? I think that would be a dandy idea with the added benefit that NVidia just upped the ante on their competitors. Any free press is GOOD PRESS! Right Lonnie? I know you're there! Make this thing happen, if you can. That will also be the "Right Thing". Ric
There are several "les"s on this list. I would like to be referred to as Lesh to help all of you keep straight who is who. I will be signing as lesh from now on.
Les Mikesell is the other Les here who is quite experienced, while I am still somewhat of a noobie, having used Linux under Fedora only about 2 years, and mostly now for reading mail and some hacking about with some 3d apps (and some toys for myself).
I don't really care where a driver comes from. FOSS would be wonderful, but I bought this computer for certain features, which I got as a real deal at the time. Now you can pretty much match it for about what I paid for it, but with inflation figured in it is now about 20% cheaper.
What I want is an OS that works efficiently (as does Linux and Windows), that is relatively safe from intrusion (Linux from my experience), and that I can understand well enough to do the coding I need to do (Windows right now, with some exceptions, and Linux) and a good support team (you guys). The rest of the discussion about why Nvidia ought to open their drivers is relatively immaterial to my personal needs.
I think the driver interface is somewhat problematic on both sides. As peripherals become smarter (to offload the cpu) the interface must evolve to deal with new capabilities, or be setup in such a way that any number of capabilities can be quickly addressed within the driver by the means of pointers to the relevant argument stream. The problem with pointers is that they give the malicious coders an open door to raid the system, and at the driver level, this is really a weakness because the requirement to "guard the gate" is very difficult when pointers are passed around.
Thus things like SELinux, virus scanners, memory gating, and so forth, all of which impact the thru-put which was the original reason for using pointers. It is a catch 22, but a vital one to address so that systems can continue to evolve.
Object programming is a potential issue of weak structure, where you don't know what was coded into the object at inception, and in many cases is not well documented. Additionally you may not be able to "figure it out" even with the code available, especially in a project of any size (thank you to the wonderful oxymoron of "self documenting code").
If we migrate to runtime objects instead of object orientation, it is possible to have malicious code that is self replicating, producing its own objects, which is another issue.
Objects also can directly contribute to code bloat. The OS's are now so huge that I don't really know how they got that big or how long they can sustain the exponential growth that has occurred. I believe object coding is part of the reason, along with some missing structural understanding that might mitigate this growth, but I haven't studied the problem in depth, because my real interests lie elsewhere, and that is the crux of the problem. No longer is it possible for one person or team capable of wrapping their hands around the full structure of an OS and the associated utilities, which further complicates the issues involved.
The proprietary driver issue is not one of simplistic issues either. If you have the code that drives a device, you have a window into the operation of that device, its strengths and weaknesses. If you are a competing vendor, you can then exploit the weaknesses for profit by building a better product. You might think the consumer wins in this case, but that is not so. This is finesse by evolution, and a group of companies will grow up focused on this aspect. Since they have no major development costs (the basics are done by their competition whose work they leverage), they can reduce their prices, to the point of driving out the innovators. The result of that becomes no "step increase" in capability, which results in only moderate change. This model will succeed, but eventually some totally new avenue will have to open for innovation to begin again, and so for some indefinite interval, the community will suffer. Think Microsoft vs Linux for this picture, on various applications.
I. The question is not whether proprietary is necessary or not, but whether we can continue innovation without proprietary profit driven innovation. Can people protect their inventions, and profit from them enough to encourage them to continue to innovate? If FOSS is dominate, how can that happen? Where will the best minds and talent go? I don't know the answer. I don't think Software should be Free as in given away, but as in freedom as the GPL is attempting to make it. I do think innovators, inventors and engineers should profit from their work, not as maintainers, but strictly on the basis of innovation and invention, and they should directly get the lions share of their innovation's rewards. Today that is not true. Maybe GPL, and Linux can change that. I hope so.
Regards, Les H
Les wrote:
Les' mission is to save us from the evils of GPL license.
There are several "les"s on this list. I would like to be referred to as Lesh to help all of
Yes, I was going to point out that I haven't been in the Navy and was probably confused with one of the others in that respect.
Regarding the GPL, though, it is all a matter of religion. Mine is that making something deliberately not interoperate with something else, whether by refusing to publish an interface spec, refusing to use standard protocols, or licensing in such a way that interoperation (or distributing working components together)is prohibited will harm random people and is thus pure evil.
Licensing in ways that have a cost per instance or per user may be moderately evil but that still lets people make their own choices based on individual merits. Taking that choice away is pure evil.
For standalone programs the GPL doesn't necessarily have these evil effects. For things that should be usable in cooperative efforts but can't because of license restrictions, it does. There's no accounting for religions, though, and no doubt others believe the harm is justified by something or other.
Les Mikesell lesmikesell@gmail.com
On Wed, 2007-10-31 at 12:56 -0500, Les Mikesell wrote:
Les wrote:
Les' mission is to save us from the evils of GPL license.
There are several "les"s on this list. I would like to be referred to as Lesh to help all of
Yes, I was going to point out that I haven't been in the Navy and was probably confused with one of the others in that respect.
---- I think that was Ricky suffering from flash back. ----
Regarding the GPL, though, it is all a matter of religion. Mine is that making something deliberately not interoperate with something else, whether by refusing to publish an interface spec, refusing to use standard protocols, or licensing in such a way that interoperation (or distributing working components together)is prohibited will harm random people and is thus pure evil.
---- you are of course calling the Linux kernel evil - notwithstanding that the intent was always to restrict the ability of commercial interests so that the source and the endless improvements upon always remained available to all users. Of course this is unlike something like a BSD license which permits absorption and further development without any requirement to release their improvements.
You are entitled to your opinion though ----
Licensing in ways that have a cost per instance or per user may be moderately evil but that still lets people make their own choices based on individual merits. Taking that choice away is pure evil.
---- Curious perspective...the only problem that I have with this is your characterization itself. ----
For standalone programs the GPL doesn't necessarily have these evil effects. For things that should be usable in cooperative efforts but can't because of license restrictions, it does. There's no accounting for religions, though, and no doubt others believe the harm is justified by something or other.
---- Seeing as how the entirety of the Linux kernel is GPL license and to change now would require a complete abandonment of the current kernel code and start from scratch, your point - however it might be made is entirely moot. The license chosen for Linux kernel development was of course Linus's and others who contribute code to the kernel are necessarily bound by the GPL license and of course, they can choose not to contribute code.
Of course the thing that makes your rich is also the thing that makes you poor and vice versa. The Linux kernel code, like all GPL license code, will always be available to continue, fork, examine, etc. and commercial enhancement of GPL code must necessarily be released in source as required...I feel rich.
Craig
Craig White wrote:
Regarding the GPL, though, it is all a matter of religion. Mine is that making something deliberately not interoperate with something else, whether by refusing to publish an interface spec, refusing to use standard protocols, or licensing in such a way that interoperation (or distributing working components together)is prohibited will harm random people and is thus pure evil.
you are of course calling the Linux kernel evil - notwithstanding that the intent was always to restrict the ability of commercial interests so that the source and the endless improvements upon always remained available to all users.
This case has never been entirely clear. In the early days, Linus stated that the kernel module interface was in fact an interface and proprietary modules were permitted. And despite current waffling, he has never clearly reversed this statement. And despite constant changes to that interface, RHEL keeps the changes from being pushed to customers within a long distro timespan, sometimes claiming both to be paying salaries to some of the kernel developers and that it is a big effort to work around the changes they are making...
Of course this is unlike something like a BSD license which permits absorption and further development without any requirement to release their improvements.
Yes, even if Microsoft cleaned up their networking by copying BSD code we are all better off with everyone using safe, well tested, standards compliant code. Apple finally produced something that could be a competitor to Windows with large chunks of it and again we are all better off for having choices instead of a monopoly.
For standalone programs the GPL doesn't necessarily have these evil effects. For things that should be usable in cooperative efforts but can't because of license restrictions, it does. There's no accounting for religions, though, and no doubt others believe the harm is justified by something or other.
Seeing as how the entirety of the Linux kernel is GPL license and to change now would require a complete abandonment of the current kernel code and start from scratch, your point - however it might be made is entirely moot.
The linux license is not quite GPL and all it would take to fix it would be for Linus to restate his original assertion that module code is not part of the kernel. Others might not agree, but he is the expert on the subject.
The license chosen for Linux kernel development was of course Linus's and others who contribute code to the kernel are necessarily bound by the GPL license and of course, they can choose not to contribute code.
And there was a reason that the kernel license contains an exception.
Of course the thing that makes your rich is also the thing that makes you poor and vice versa. The Linux kernel code, like all GPL license code, will always be available to continue, fork, examine, etc. and commercial enhancement of GPL code must necessarily be released in source as required...I feel rich.
I think opensolaris has potential as an alternative, especially with distributions like nexenta that have fairly current userland programs equivalent to ubuntu on top of it.
On Wed, 2007-10-31 at 15:41 -0500, Les Mikesell wrote:
Craig White wrote:
Of course this is unlike something like a BSD license which permits absorption and further development without any requirement to release their improvements.
Yes, even if Microsoft cleaned up their networking by copying BSD code we are all better off with everyone using safe, well tested, standards compliant code. Apple finally produced something that could be a competitor to Windows with large chunks of it and again we are all better off for having choices instead of a monopoly.
---- Apple and Microsoft...two peas in a pod, remind me, what did Billy Martin say about George and Reggie?
s/monopoly/liar
;-)
Craig
On Wed, 2007-10-31 at 08:14 +0000, Andy Green wrote:
Somebody in the thread at some point said:
His position is the same as mine that if the GPL stops you **utterly in your tracks** from doing useful stuff, then the old-school "Right Thing" principle kicks in. You look for a solution, any solution to get the job done. Something is rotten in Denmark when the goal is being righteous as
The GPL doesn't stop you combining stuff, it stops you distributing some combinations. That's a bit different. Proprietary licensed stuff is far worse, you can't redistribute it AT ALL or you get chucked in jail.
Please explain. If nVidia states on their site that it's perfectly ok to redistribute their driver package, then it's Richard Stallman that is going to chuck me in jail? <cackles hugely and smirks>
This and his unwillingness to converge on any agreement different from his starting position is my problem with Les' complaints, despite I have seen him say a lot of smart things otherwise.
All I have gathered is that his stance is that it shouldn't be a show stopper until a better solution is found. When I worked in Installation Support, back when, it was not uncommon for us to recommend OSS to enable a users sound card. While Hannu did a bunch of support for a free OSS for Linux, the version he sold for a mere $10 was better. Oddball sound cards (mainly the on-board variety) became usable and people were happy. We had our queues cleared out of sound card problems too, nor did anyone every complain about the GPL. The end-user was enjoined to not redistribute the commercial version of OSS.
Sometimes you have to use proprietary stuff to do what you need to do at all -- but it wasn't that long ago that 100% proprietary was the only choice. Now for a lot of us it is the unusual case at the edge on our PCs and even outside of PCs FOSS solutions are creeping into the other devices around us too.
It probably always will. Compromise. A wonderful and beautiful word. Perspective, also a word that you'll probably never hear used enough by those in Power & Authority (or by those that want to be in that position).
But choosing to use a proprietary solution as the enduser is different from changing the basis of a Free distro to include it. Whereas we individually might have to compromise to go on, the ideal shouldn't be compromised (he said, trying not to think about firmware exceptions).
-Andy