8:22 p.m.
We are all being shafted in this one. For those of you like me that have
experienced high quality video courtesy of PWCX.....
http://www.smcc.demon.nl/webcam/
Is their any way to talk some sense into the Kernel maintainers?
What is next, don't allow nvidia driver to load?
Ted
8:49 p.m.
On Sat, 2004-08-28 at 21:22 -0400, Listman wrote:
We are all being shafted in this one. For those of you like me that
have
experienced high quality video courtesy of PWCX.....
http://www.smcc.demon.nl/webcam/
Is their any way to talk some sense into the Kernel maintainers?
You might want to read Greg K-H's faq before you go 'talking sense' into
anyone.
http://lwn.net/Articles/99639/
What is next, don't allow nvidia driver to load?
This would make an enormous number of fedora devel people happy. :)
-sv
8:51 p.m.
On Sat, 2004-08-28 at 21:49 -0400, seth vidal wrote:
On Sat, 2004-08-28 at 21:22 -0400, Listman wrote:
> What is next, don't allow nvidia driver to load?
>
This would make an enormous number of fedora devel people happy. :)
-sv
FWIW, it would mean (at least) one less fedora devel guinea pig.
--
Aaron Gaudio <prothonotar(a)tarnation.dyndns.org>
9 p.m.
FWIW, it would mean (at least) one less fedora devel guinea pig.
The biggest problem with the nvidia drivers is that when something
breaks, and a kernel or X devel person sees that the user is using the
nvidia drivers, then they're stuck. They can't go look through code and
find the problem. They come up to the binary-only nvidia driver and have
to stop.
That's a whole world of no-fun.
And the sad reality is, lots of people use these drivers and almost all
of them post about bugs in X or at the kernel level with these drivers
loaded.
-sv
4:23 a.m.
On Sat, 2004-08-28 at 22:00 -0400, seth vidal wrote:
> FWIW, it would mean (at least) one less fedora devel guinea
pig.
The biggest problem with the nvidia drivers is that when something
breaks, and a kernel or X devel person sees that the user is using the
nvidia drivers, then they're stuck. They can't go look through code and
find the problem. They come up to the binary-only nvidia driver and have
to stop.
That's a whole world of no-fun.
And the sad reality is, lots of people use these drivers and almost all
of them post about bugs in X or at the kernel level with these drivers
loaded.
Yeah, I can understand the frustration at this. I agree that if
someone's having a problem with an binary-only driver, their only
recourse is the provider of that driver.
--
Aaron Gaudio <prothonotar(a)tarnation.dyndns.org>
4:40 a.m.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Sunday 29 August 2004 10:23, Aaron Gaudio wrote:
On Sat, 2004-08-28 at 22:00 -0400, seth vidal wrote:
> > FWIW, it would mean (at least) one less fedora devel guinea pig.
> And the sad reality is, lots of people use these drivers and
almost all
> of them post about bugs in X or at the kernel level with these drivers
> loaded.
Yeah, I can understand the frustration at this. I agree that if
someone's having a problem with an binary-only driver, their only
recourse is the provider of that driver.
One should also note that the act of signing the NDA put the author in this
bad spot from the get-go. He can't really please both "masters" at that
point.
The membrane between "you must GPL your stuff" and "keep your source"
is thin
at the kernel modules. I can imagine why backdoor hooks to let Philips "off
the hook" were considered evil since widespread use of them would allow the
GPLness fo the kernel to be completely worked around.
Anyway, if someone else takes over maintainership they will be free to reverse
the binary portion and solve the problem that way. If not, time for a new
webcam.
- -Andy
- --
Automatic actions for USB cameras, cardreaders, memory sticks, MP3 players
http://warmcat.com/usbautocam
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)
iD8DBQFBMaSpjKeDCxMJCTIRAjaZAKCYKNcbPzX9KPV9di5m/ylW603prACfY4+0
3u3m/oRoxhIah+Arl/8QfS8=
=JUfa
-----END PGP SIGNATURE-----
1:29 p.m.
On Sun, Aug 29, 2004 at 10:40:53 +0100,
Andy Green <fedora(a)warmcat.com> wrote:
One should also note that the act of signing the NDA put the author in this
bad spot from the get-go. He can't really please both "masters" at that
point.
The NDA appears to have expired over a year ago.
The membrane between "you must GPL your stuff" and
"keep your source" is thin
at the kernel modules. I can imagine why backdoor hooks to let Philips "off
the hook" were considered evil since widespread use of them would allow the
GPLness fo the kernel to be completely worked around.
Anyway, if someone else takes over maintainership they will be free to reverse
the binary portion and solve the problem that way. If not, time for a new
webcam.
The discussion of the binary has been misleading as well. It isn't primarily
a decompressor, but rather an interpolator. The ccd is 120x160 pixels
and the binary part interperolates to return 480x640 pixels. I think it
is pretty questionable that that function should be done in kernel space.
1:30 p.m.
On Tue, Aug 31, 2004 at 01:29:43PM -0500, Bruno Wolff III wrote:
The discussion of the binary has been misleading as well. It
isn't primarily
a decompressor, but rather an interpolator. The ccd is 120x160 pixels
and the binary part interperolates to return 480x640 pixels. I think it
is pretty questionable that that function should be done in kernel space.
Its a 640x480 detector according to the component data sheet so I wouldn't
read much into random rumour. It looks like the matter is now resolved
anyway and we may also get open user space decode libraries
9:08 p.m.
On Sat, 2004-08-28 at 21:51, Aaron Gaudio wrote:
On Sat, 2004-08-28 at 21:49 -0400, seth vidal wrote:
> On Sat, 2004-08-28 at 21:22 -0400, Listman wrote:
> > What is next, don't allow nvidia driver to load?
> >
>
> This would make an enormous number of fedora devel people happy. :)
> -sv
FWIW, it would mean (at least) one less fedora devel guinea pig.
--
Aaron Gaudio <prothonotar(a)tarnation.dyndns.org>
Fact's are fact's, who suffer's here?
Aren't we the "community" trying to promote linux?
Look, no one in their right mind can argue that anything binary only is
good, but anyone using Open Source who installs a binary at their own
risk. And I am certain much good code has come back from the vendors to
the community in response to getting their binary only stuff working.
As far as fedora devel being happy about not allowing nvidia to load let
me know, I will but some stock in ATI :-)
Maybe I am way off base, and the amount of deployed Linux is so huge
that it can tell all the Vendors as well as Hardware engineers in the
world to go screw, its our way or the highway.
Wake up, the rainbow has risen, but still quite a few colors are
missing.
Ted
10:05 p.m.
seth vidal wrote:
On Sat, 2004-08-28 at 21:22 -0400, Listman wrote:
>We are all being shafted in this one. For those of you like me that have
>experienced high quality video courtesy of PWCX.....
>
>http://www.smcc.demon.nl/webcam/
>
>Is their any way to talk some sense into the Kernel maintainers?
>
You might want to read Greg K-H's faq before you go 'talking sense' into
anyone.
http://lwn.net/Articles/99639/
For what it's worth, my view is that Nemosoft is right.
Let me declare my interest, I have two of these webcams - a Philips
ToUcam Pro (PCVC 740K) and
a similar Logictech webcam. I bought these after some research on the
web because there was support
for them and because they were high quality with CCD sensors. The
Logictech device (or its successor)
is widely available at PC World, and my philips cam is in use on my
website at http://80.4.116.219/cam.html
(in the daylight hours anyway) on an earlier version of redhat linux
(7.3.94).
I have noticed that these cams do not work on Fedora - and this inhibits
me from moving my site to Fedora
as I had planned to do. Never mind the religeous aspects of kernel
tainting, what matters to me is whether
I can use the system.
I get a feeling too that there is an effort at defeaturing "Linux". I
believe that vsftpd doesn't now work in
passive mode (on FC2)- at least without strong linux admin skills - so
it appears that in certain areas, using
linux is becoming harder. Whether this is deliberate or not I cannot say.
10:18 p.m.
seth vidal wrote:
On Sat, 2004-08-28 at 21:22 -0400, Listman wrote:
>We are all being shafted in this one. For those of you like me that have
>experienced high quality video courtesy of PWCX.....
>
>http://www.smcc.demon.nl/webcam/
>
>Is their any way to talk some sense into the Kernel maintainers?
>
You might want to read Greg K-H's faq before you go 'talking sense' into
anyone.
http://lwn.net/Articles/99639/
>What is next, don't allow nvidia driver to load?
>
This would make an enormous number of fedora devel people happy. :)
-sv
Thanks for the link to the articles. I especially enjoyed the comment
regarding legality and morality.
And I thought that most of these lists were pretty tame and dry.
On the inclusion of hooks and binary only drivers. It is understandable
that the source code is freely available, hooks can be added. A user
space alternative can be come up with to overcome the removal of the
hook from the kernel.
On the hardware support for as many devices that are possible,
especially issues related to getting highly functional and useful
devices running optimally, it would be decent to have.
Of course, I'm carrying a binary driver around recently myself
(i810_dvr.o) until the development build catches up with my hardware
used. I know the patch and rebuilding the package from source would be
ideal. I find that it is easier than installing xorg-x11 to test to see
if it works, finding out that it doesn't work, then reinstalling all of
the older rpms or recompiling xorg-x11 from source with each new release.
Basically, I see the benifits from an all open source system and would
ideally shoot for this sysem. I also understand that sometimes being too
pure with the source vs. binary issues can be time consuming and a
challenge on those that are just shooting for a great running and
hardware richly supported system that doesn't rip you off by closing the
code to the world.
Nvidia and the PWCX can move to open source and reduce their overhead
and allow others to improve the software that makes their hardware
function correctly. I'm sure that the respect for the industries would
be recognized and they would have input from others on how to optimize
their hardware functionality even more.
my morality vs. legality view,
Jim
--
Linus, Alan - Please apply the following self-explanatory patch.
+ /* LynuxWorks are politely reminded that removing copyright
+ notices is an offence under the Copyright Design and
+ Patents Act 1988, and under equivalent non-UK law in
+ accordance with the Berne Convention. */
+ printk("Portions (C) 2000, 2001 Red Hat, Inc.\n");
- David Woodhouse on linux-kernel
5:12 a.m.
On Sun, 2004-08-29 at 03:22, Listman wrote:
We are all being shafted in this one. For those of you like me that
have
experienced high quality video courtesy of PWCX.....
http://www.smcc.demon.nl/webcam/
Is their any way to talk some sense into the Kernel maintainers?
Is there any way to talk some sense into all these people who don't even
take the time to understand the issue before moaning loudly?
I mean; the guy demanded his driver to be removed from Linus' tree. We
might just put the driver back for fedora. We don't know yet.
As for the hook thing and binary modules... Personally I consider about
all the binary only kernel modules that are distributed in violation of
the GPL. I'm not saying one cannot do one which doesn't, just that I
haven't seen many that don't. Note that that is my personal opinion and
not my employers'. Nor am I a lawyer.
Linus and the bulk of the kernel developers want the kernel to be free
of pure hooks for binary modules regardless of the outcome of the
gray-area discussion. The PWC driver author somehow doesn't agree with
that policy and no longer wants to maintain his driver and has requested
his driver to be removed. His right to ask that.... Others have been
reverse engineering this thing in the last few days and it appears it's
just a fancy interpolation scheme... quite valuable IP that is ;)
What most "uninformed" people who moan on several mailinglists seem to
ignore is that the GPL offers anyone the freedom to keep using the
driver. Linus did not say "Nobody can use this driver". He said "I
respect the wishes of the author to remove this". Just like Greg/Linus
removing the hook in their trees doesn't mean you or anyone else doesn't
have the freedom to put them back. The only snag is that the nemosoft
guy no longer distributes the binary only bits... which is the part that
isn't covered by the freedom of the GPL; anyone depending on those bits
voluntarily gave up their freedom and I find it really hard to feel
pitty for that.
The apparent PR campaign (including your email) actually does more
damage than good; all these random emails contain no new arguments other
than repeating the incorrect assertions, and as such make the people
that would matter actually care *less* not more about this issue.
9:06 a.m.
On Sun, 2004-08-29 at 06:12, Arjan van de Ven wrote:
On Sun, 2004-08-29 at 03:22, Listman wrote:
> We are all being shafted in this one. For those of you like me that have
> experienced high quality video courtesy of PWCX.....
>
> http://www.smcc.demon.nl/webcam/
>
> Is their any way to talk some sense into the Kernel maintainers?
Is there any way to talk some sense into all these people who don't even
take the time to understand the issue before moaning loudly?
I mean; the guy demanded his driver to be removed from Linus' tree. We
might just put the driver back for fedora. We don't know yet.
As for the hook thing and binary modules... Personally I consider about
all the binary only kernel modules that are distributed in violation of
the GPL. I'm not saying one cannot do one which doesn't, just that I
haven't seen many that don't. Note that that is my personal opinion and
not my employers'. Nor am I a lawyer.
Linus and the bulk of the kernel developers want the kernel to be free
of pure hooks for binary modules regardless of the outcome of the
gray-area discussion. The PWC driver author somehow doesn't agree with
that policy and no longer wants to maintain his driver and has requested
his driver to be removed. His right to ask that.... Others have been
reverse engineering this thing in the last few days and it appears it's
just a fancy interpolation scheme... quite valuable IP that is ;)
What most "uninformed" people who moan on several mailinglists seem to
ignore is that the GPL offers anyone the freedom to keep using the
driver. Linus did not say "Nobody can use this driver". He said "I
respect the wishes of the author to remove this". Just like Greg/Linus
removing the hook in their trees doesn't mean you or anyone else doesn't
have the freedom to put them back. The only snag is that the nemosoft
guy no longer distributes the binary only bits... which is the part that
isn't covered by the freedom of the GPL; anyone depending on those bits
voluntarily gave up their freedom and I find it really hard to feel
pitty for that.
The apparent PR campaign (including your email) actually does more
damage than good; all these random emails contain no new arguments other
than repeating the incorrect assertions, and as such make the people
that would matter actually care *less* not more about this issue.
After reflecting on this, I have to agree. What is the point in wasting
core resources? Near term benefits at the cost of long term gains are
always bad under almost all circumstances and especially in this one.
If anything, the user base should be the ones fighting these battles
with the vendors, not the developers.
Kicks self in @$$.
Ted
11:05 a.m.
New subject: OffList: Whats with the petty Open Source Pureness that has killed PWCX
On Sun, 29 Aug 2004 12:12:54 +0200, Arjan van de Ven
<arjan(a)fenrus.demon.nl> wrote:
I mean; the guy demanded his driver to be removed from Linus'
tree. We
might just put the driver back for fedora. We don't know yet.
I've taken this off-list because i don't want my suggestion to be a
launching point for Axel or someone for ranting.
If you decide to re-include this driver in fedora core even though its
not in the mainline tree, I would like to suggest that this driver be
maintained as a seperate add-on module package inside Core and not as
a patch to the kernel. I'm sure doing that has some down-sides in
terms of package maintainance, but i think having at least one example
of a seperate kernel module package in Core would be useful as a some
what authoritive reference to people creating add-on kernel modules in
Extras.
Either way its just a suggestion, in the hopes of making some headway
in the never ending kernel-modules packaging issues.
-jef
6:15 p.m.
New subject: OffList: Whats with the petty Open Source Pureness that has killed PWCX
On Sun, Aug 29, 2004 at 12:05:08PM -0400, Jeff Spaleta wrote:
On Sun, 29 Aug 2004 12:12:54 +0200, Arjan van de Ven
<arjan(a)fenrus.demon.nl> wrote:
> I mean; the guy demanded his driver to be removed from Linus' tree. We
> might just put the driver back for fedora. We don't know yet.
I've taken this off-list because i don't want my suggestion to be a
launching point for Axel or someone for ranting.
Looks like it's on-list, but I'll keep from ranting this time ;)
If you decide to re-include this driver in fedora core even though
its
not in the mainline tree, I would like to suggest that this driver be
maintained as a seperate add-on module package inside Core and not as
a patch to the kernel. I'm sure doing that has some down-sides in
terms of package maintainance, but i think having at least one example
of a seperate kernel module package in Core would be useful as a some
what authoritive reference to people creating add-on kernel modules in
Extras.
Either way its just a suggestion, in the hopes of making some headway
in the never ending kernel-modules packaging issues.
--
Axel.Thimm at ATrpms.net
10 p.m.
New subject: OffList: Whats with the petty Open Source Pureness that has killed PWCX
On Sun, 29 Aug 2004, Jeff Spaleta wrote:
I've taken this off-list because i don't want my suggestion
to be a
launching point for Axel or someone for ranting.
Except it went on-list.
List maintainers: setting Reply-To list is *evil*, please consider
removing this malfeature of the RedHat lists. Luckily for Jeff his
mail wasnt sensitive..
regards,
--
Paul Jakma paul(a)clubi.ie paul(a)jakma.org Key ID: 64A2FF6A
Fortune:
Anarchy may not be a better form of government, but it's better than no
government at all.
10:08 p.m.
New subject: OffList: Whats with the petty Open Source Pureness that has killed PWCX
On Mon, 2004-08-30 at 04:00 +0100, Paul Jakma wrote:
On Sun, 29 Aug 2004, Jeff Spaleta wrote:
> I've taken this off-list because i don't want my suggestion to be a
> launching point for Axel or someone for ranting.
Except it went on-list.
List maintainers: setting Reply-To list is *evil*, please consider
removing this malfeature of the RedHat lists. Luckily for Jeff his
mail wasnt sensitive..
List maintainers, consider pruning all mails that discuss the sinfulness
(or not) of Reply-To as it is an excellent way to make lists more
annoying. :)
-sv
1:21 p.m.
New subject: OffList: Whats with the petty Open Source Pureness that has killed PWCX
On Mon, Aug 30, 2004 at 04:00:39 +0100,
Paul Jakma <paul(a)dishone.st> wrote:
On Sun, 29 Aug 2004, Jeff Spaleta wrote:
List maintainers: setting Reply-To list is *evil*, please consider
removing this malfeature of the RedHat lists. Luckily for Jeff his
mail wasnt sensitive..
A useful defense against this is to have a mail filter remove reply-to
headers from lists that you know to do this.
9:40 p.m.
On Sat, 28 Aug 2004, Listman wrote:
We are all being shafted in this one. For those of you like me that
have
experienced high quality video courtesy of PWCX.....
What is next, don't allow nvidia driver to load?
Not at all. But expecting closed source drivers to get distributed in the
official kernel is silly. Properly understood your question inverts to
"Why doesn't Linus include the Nvidia module with the kernel?" and the
answer is obvious. And yes it is logically equivilent since the kernel
module for the NVidia driver is open source (or was back when I bought
Nvidia's closed crap.) but is only useful to interface to the closed
source X module. The pwcx driver was pretty much useless (160x120 max
res) without the closed source addon so should NOT be in the mainline
kernel. Nothing would have stopped the maintainer from offering the
closed source pwcx driver as a standalone module just like the winmodem
drivers, the ati & nvidia video drivers, the vmware kernel modules, etc.
But if you have been following along, yes recent Fedora (FC2's shipping
package) kernels DIDN'T allow the Nvidia driver to load because the
underlying internals had changed. Because nobody in the Free Software
World really cares about closed drivers, because we don't use them.
Nvidia eventually caught up and again, nobody actually writing code cares.
Closed software is, for the most part, something that happens "Somewhere
else."
Me, I ain't quite as orthodox as the RMS on the issue, I avoid closed
drivers like the plague but when all else fails I will use one.
Winmodems for example, since there are no Free drivers and due to patents
probably never will be.
--
John M. http://www.beau.org/~jmorris This post is 100% M$ Free!
Geekcode 3.1:GCS C+++ UL++++$ P++ L+++ W++ w--- Y++ b++ 5+++ R tv- e* r
7178
days inactive
7180
days old
18 comments
13 participants
participants (13)
-
Aaron Gaudio -
Alan Cox -
Andy Green -
Arjan van de Ven -
Axel Thimm -
Bruno Wolff III -
Jef Spaleta -
Jim Cornette -
John Morris -
Listman -
Paul Jakma -
richard mullens -
seth vidal