4:09 p.m.
Hi
Doing a harddisk installation of the Fedora 7 Test 2 live CD ended up
retaining the liveinst package which also shows up a menu entry "install
to harddisk" even after the hard disk installation is complete. Running
that exits with a not a live cd error however. On a installed system
this menu entry should be hidden somehow.
The latest update of this system today results in
Transaction Check Error:
file /usr/sbin/liveinst from install of anaconda-11.2.0.40-1
conflicts with file from package liveinst-0.01-1
file /usr/share/applications/liveinst.desktop from install of
anaconda-11.2.0.40-1 conflicts from package liveinst-0.01-1
Rahul
4:15 p.m.
Am Sat, 24 Mar 2007 02:39:37 +0530
schrieb Rahul Sundaram <sundaram(a)fedoraproject.org>:
Hi
Doing a harddisk installation of the Fedora 7 Test 2 live CD ended up
retaining the liveinst package which also shows up a menu entry
"install to harddisk" even after the hard disk installation is
complete. Running that exits with a not a live cd error however. On
a installed system this menu entry should be hidden somehow.
The latest update of this system today results in
Transaction Check Error:
file /usr/sbin/liveinst from install of anaconda-11.2.0.40-1
conflicts with file from package liveinst-0.01-1
file /usr/share/applications/liveinst.desktop from install of
anaconda-11.2.0.40-1 conflicts from package liveinst-0.01-1
The liveinst bits are included in anaconda since anaconda-11.2.0.39-1.
So liveinst packagae is not needed anymore.
Sebastian
4:19 p.m.
Sebastian Vahl wrote:
Am Sat, 24 Mar 2007 02:39:37 +0530
schrieb Rahul Sundaram <sundaram(a)fedoraproject.org>:
> Hi
>
> Doing a harddisk installation of the Fedora 7 Test 2 live CD ended up
> retaining the liveinst package which also shows up a menu entry
> "install to harddisk" even after the hard disk installation is
> complete. Running that exits with a not a live cd error however. On
> a installed system this menu entry should be hidden somehow.
>
> The latest update of this system today results in
>
> Transaction Check Error:
> file /usr/sbin/liveinst from install of anaconda-11.2.0.40-1
> conflicts with file from package liveinst-0.01-1
> file /usr/share/applications/liveinst.desktop from install of
> anaconda-11.2.0.40-1 conflicts from package liveinst-0.01-1
>
The liveinst bits are included in anaconda since anaconda-11.2.0.39-1.
So liveinst packagae is not needed anymore.
I know that but then Anaconda should have obsoleted liveinst to ensure a
proper update path. That still leaves the misleading install to hard
disk menu entry on a installed system.
Rahul
4:24 p.m.
On Sat, 2007-03-24 at 02:49 +0530, Rahul Sundaram wrote:
I know that but then Anaconda should have obsoleted liveinst to
ensure a
proper update path. That still leaves the misleading install to hard
disk menu entry on a installed system.
Why would we add an obsolete for a package that was never in any
official repository?
Jeremy
4:29 p.m.
Jeremy Katz wrote:
On Sat, 2007-03-24 at 02:49 +0530, Rahul Sundaram wrote:
> I know that but then Anaconda should have obsoleted liveinst to ensure a
> proper update path. That still leaves the misleading install to hard
> disk menu entry on a installed system.
Why would we add an obsolete for a package that was never in any
official repository?
It was in the Fedora 7 Test 2 live cd. I got it when I installed it to
my hard disk. Doing a yum update today results in a conflict. Why should
a end user run into this?
Rahul
4:39 p.m.
On Friday 23 March 2007 17:29:37 Rahul Sundaram wrote:
It was in the Fedora 7 Test 2 live cd. I got it when I installed it
to
my hard disk. Doing a yum update today results in a conflict. Why should
a end user run into this?
Trying to expect sane paths from anything less then Test3 is pretty
unreasonable.
--
Jesse Keating
Release Engineer: Fedora
4:49 p.m.
Jesse Keating wrote:
On Friday 23 March 2007 17:29:37 Rahul Sundaram wrote:
> It was in the Fedora 7 Test 2 live cd. I got it when I installed it to
> my hard disk. Doing a yum update today results in a conflict. Why should
> a end user run into this?
Trying to expect sane paths from anything less then Test3 is pretty
unreasonable.
It is not unreasonable at all. The liveinst package should have been on
the officially repositories. The review and the process was bypassed to
get test 2 on time which is fine but let's not use that as a excuse not
to add a simple obsolete to Anaconda to ensure that users can update
properly. We should strive to provide sane updates paths from every
release to every release regardless of whether it is a test or not. It
might not be possible in some corner cases which this is not. Our end
users and testers deserve more respect for their time.
Rahul
4:53 p.m.
On Sat, 24 Mar 2007, Rahul Sundaram wrote:
It is not unreasonable at all. The liveinst package should have been
on the
officially repositories. The review and the process was bypassed to get test
2 on time which is fine but let's not use that as a excuse not to add a
simple obsolete to Anaconda to ensure that users can update properly. We
should strive to provide sane updates paths from every release to every
release regardless of whether it is a test or not. It might not be possible
in some corner cases which this is not. Our end users and testers deserve
more respect for their time.
Assuming there are no unintended consequence for adding this obsolete, and
assuming that it will make a better experience for some srt of users, and
assuming that it's a simple fix...
...I find myself in agreement with Rahul. But then, that's a lot of
assumptions, and I'm no expert, eh?
--g
-------------------------------------------------------------
Greg DeKoenigsberg || Fedora Project || fedoraproject.org
Be an Ambassador || http://fedoraproject.org/wiki/Ambassadors
-------------------------------------------------------------
5 p.m.
On Sat, 2007-03-24 at 03:19 +0530, Rahul Sundaram wrote:
Jesse Keating wrote:
> On Friday 23 March 2007 17:29:37 Rahul Sundaram wrote:
>> It was in the Fedora 7 Test 2 live cd. I got it when I installed it to
>> my hard disk. Doing a yum update today results in a conflict. Why should
>> a end user run into this?
>
> Trying to expect sane paths from anything less then Test3 is pretty
> unreasonable.
It is not unreasonable at all. The liveinst package should have been on
the officially repositories. The review and the process was bypassed to
get test 2 on time which is fine but let's not use that as a excuse not
to add a simple obsolete to Anaconda to ensure that users can update
properly. We should strive to provide sane updates paths from every
release to every release regardless of whether it is a test or not. It
might not be possible in some corner cases which this is not. Our end
users and testers deserve more respect for their time.
Or maybe the developers deserve more of your respect for their time?
The end users will never see this; the testers know to expect this sort
of thing. Anyone who expects upgrades between experimental test releases
to work flawlessly has got their expectations seriously out of whack.
Here's the deal: We've committed to do the best we can to ensure that
upgrades between major releases work and that Test3 (or, in this case,
Test4)->final upgrades will work. Everything beyond that is
best-effort.
In my opinion this can be handled just fine by a Test3 release note.
-w
5:17 p.m.
Will Woods wrote:
On Sat, 2007-03-24 at 03:19 +0530, Rahul Sundaram wrote:
> Jesse Keating wrote:
>> On Friday 23 March 2007 17:29:37 Rahul Sundaram wrote:
>>> It was in the Fedora 7 Test 2 live cd. I got it when I installed it to
>>> my hard disk. Doing a yum update today results in a conflict. Why should
>>> a end user run into this?
>> Trying to expect sane paths from anything less then Test3 is pretty
>> unreasonable.
> It is not unreasonable at all. The liveinst package should have been on
> the officially repositories. The review and the process was bypassed to
> get test 2 on time which is fine but let's not use that as a excuse not
> to add a simple obsolete to Anaconda to ensure that users can update
> properly. We should strive to provide sane updates paths from every
> release to every release regardless of whether it is a test or not. It
> might not be possible in some corner cases which this is not. Our end
> users and testers deserve more respect for their time.
Or maybe the developers deserve more of your respect for their time?
Maybe if you have less users than developers (which is thankfully not
the case as you can see from our statistics page). Otherwise no. See,
this is a very very trivial change and could have been done in maybe 2
seconds. Why are we spending so much time arguing about it? Are
developers here somehow philosophically opposed to adding a obsolete to
a package to make it easier for testers to update and provide feedback
to the project?
The end users will never see this; the testers know to expect this
sort
of thing. Anyone who expects upgrades between experimental test releases
to work flawlessly has got their expectations seriously out of whack.
I am not expecting everything to work flawlessly. I am asking that we do
simple things to make it easier for end users to continue updating and
providing feedback to us.
Here's the deal: We've committed to do the best we can to
ensure that
upgrades between major releases work and that Test3 (or, in this case,
Test4)->final upgrades will work. Everything beyond that is
best-effort.
In my opinion this can be handled just fine by a Test3 release note.
As someone who argued in the Fedora QA meeting to bring in this change I
consider ensuring upgrades from test 3 to final release as a good first
step. We could do better in the future. There are other distributions
like Debian which have paved the way. It is very critical especially for
Fedora to do this because it brings in more early testers.
Have you wondered why upstream projects like GNOME which Fedora has very
much been aligned with dont see much bug reports from Fedora rawhide
users or why our general releases have always atleast one major
embrassing bug? Ensuring proper update paths from even test releases
will definitely improve the quality of the general releases.
Rahul
6250
days inactive
6250
days old
livecd@lists.fedoraproject.org
9 comments
6 participants
participants (6)
-
Greg DeKoenigsberg -
Jeremy Katz -
Jesse Keating -
Rahul Sundaram -
Sebastian Vahl -
Will Woods