Sorry for the overdramatic subject, but this is important.
Within the next few days Python 2.6 will be imported into Rawhide. This means that EVERY single Python-based package in Rawhide will be broken, and that we'll need to slog our way through rebuilding it package by package. I'm going to forgo the customary list of packages this time since they number in the hundreds. You can see them for yourself by using the following:
repoquery --disablerepo=* --enablerepo={development,rawhide} \ --whatrequires "python(abi)" | sort | less
Good luck, and godspeed.
On Mon, 24 Nov 2008, Ignacio Vazquez-Abrams wrote:
Sorry for the overdramatic subject, but this is important.
Within the next few days Python 2.6 will be imported into Rawhide. This means that EVERY single Python-based package in Rawhide will be broken, and that we'll need to slog our way through rebuilding it package by package. I'm going to forgo the customary list of packages this time since they number in the hundreds. You can see them for yourself by using the following:
repoquery --disablerepo=* --enablerepo={development,rawhide} \ --whatrequires "python(abi)" | sort | less
Good luck, and godspeed.
So, unless everyone is comfy with rawhide being broken for most of the rest of the week, I'd suggest we delay the import of python 2.6 until after the weekend and a huge portion of the americans in working on fedora get back from thanksgiving food comas.
-sv
Seth Vidal (skvidal@fedoraproject.org) said:
So, unless everyone is comfy with rawhide being broken for most of the rest of the week, I'd suggest we delay the import of python 2.6 until after the weekend and a huge portion of the americans in working on fedora get back from thanksgiving food comas.
I'm taking it from the original mail that the python maintainers do not want to trigger the rebuilds?
Bill
On Mon, 2008-11-24 at 13:09 -0500, Bill Nottingham wrote:
Seth Vidal (skvidal@fedoraproject.org) said:
So, unless everyone is comfy with rawhide being broken for most of the rest of the week, I'd suggest we delay the import of python 2.6 until after the weekend and a huge portion of the americans in working on fedora get back from thanksgiving food comas.
I'm taking it from the original mail that the python maintainers do not want to trigger the rebuilds?
I don't mind doing some of them, but there are almost 700 total and there's no way I can get that many done in a timely fashion.
On Mon, 2008-11-24 at 13:16 -0500, Ignacio Vazquez-Abrams wrote:
On Mon, 2008-11-24 at 13:09 -0500, Bill Nottingham wrote:
Seth Vidal (skvidal@fedoraproject.org) said:
So, unless everyone is comfy with rawhide being broken for most of the rest of the week, I'd suggest we delay the import of python 2.6 until after the weekend and a huge portion of the americans in working on fedora get back from thanksgiving food comas.
I'm taking it from the original mail that the python maintainers do not want to trigger the rebuilds?
I don't mind doing some of them, but there are almost 700 total and there's no way I can get that many done in a timely fashion.
You might consider a separate koji tag for this effort, similar to what I did for perl in the F9 timeframe.
~spot
On Mon, 2008-11-24 at 13:18 -0500, Tom "spot" Callaway wrote:
On Mon, 2008-11-24 at 13:16 -0500, Ignacio Vazquez-Abrams wrote:
On Mon, 2008-11-24 at 13:09 -0500, Bill Nottingham wrote:
Seth Vidal (skvidal@fedoraproject.org) said:
So, unless everyone is comfy with rawhide being broken for most of the rest of the week, I'd suggest we delay the import of python 2.6 until after the weekend and a huge portion of the americans in working on fedora get back from thanksgiving food comas.
I'm taking it from the original mail that the python maintainers do not want to trigger the rebuilds?
I don't mind doing some of them, but there are almost 700 total and there's no way I can get that many done in a timely fashion.
You might consider a separate koji tag for this effort, similar to what I did for perl in the F9 timeframe.
It's also what we've done for python in the past. It's good to at least get the stack up through yum and friends building and working before thrusting the new python upon everyone as otherwise it's quite difficult for people to even try to fix things on their own.
Jeremy
On Mon, 24 Nov 2008, Jeremy Katz wrote:
It's also what we've done for python in the past. It's good to at least get the stack up through yum and friends building and working before thrusting the new python upon everyone as otherwise it's quite difficult for people to even try to fix things on their own.
So we'll need:
- pygpgme - python-iniparse - python-urlgrabber - rpm-python - yum-metadata-parser (python module)
-sv
On Mon, Nov 24, 2008 at 1:33 PM, Seth Vidal skvidal@fedoraproject.org wrote:
On Mon, 24 Nov 2008, Jeremy Katz wrote:
It's also what we've done for python in the past. It's good to at least get the stack up through yum and friends building and working before thrusting the new python upon everyone as otherwise it's quite difficult for people to even try to fix things on their own.
So we'll need:
- pygpgme
- python-iniparse
- python-urlgrabber
- rpm-python
- yum-metadata-parser (python module)
python-setuptools will need to work for most other python packages to build.
Cheers,
On Mon, 24 Nov 2008, Konstantin Ryabitsev wrote:
On Mon, Nov 24, 2008 at 1:33 PM, Seth Vidal skvidal@fedoraproject.org wrote:
On Mon, 24 Nov 2008, Jeremy Katz wrote:
It's also what we've done for python in the past. It's good to at least get the stack up through yum and friends building and working before thrusting the new python upon everyone as otherwise it's quite difficult for people to even try to fix things on their own.
So we'll need:
- pygpgme
- python-iniparse
- python-urlgrabber
- rpm-python
- yum-metadata-parser (python module)
python-setuptools will need to work for most other python packages to build.
Good point. Shouldn't be too much beyond that, though.
-sv
On Mon, 2008-11-24 at 13:33 -0500, Seth Vidal wrote:
On Mon, 24 Nov 2008, Jeremy Katz wrote:
It's also what we've done for python in the past. It's good to at least get the stack up through yum and friends building and working before thrusting the new python upon everyone as otherwise it's quite difficult for people to even try to fix things on their own.
So we'll need:
- pygpgme
- python-iniparse
- python-urlgrabber
- rpm-python
- yum-metadata-parser (python module)
I got all those but python-iniparse (and y-m-p, which I didn't try) built locally, and it only failed because of local reasons (i.e., I didn't bump the release, I just rebuilt the SRPM as-is).
On Mon, 2008-11-24 at 13:18 -0500, Tom "spot" Callaway wrote:
You might consider a separate koji tag for this effort, similar to what I did for perl in the F9 timeframe.
I can do that. Just point me in the right direction please.
On Mon, 2008-11-24 at 13:41 -0500, Ignacio Vazquez-Abrams wrote:
On Mon, 2008-11-24 at 13:18 -0500, Tom "spot" Callaway wrote:
You might consider a separate koji tag for this effort, similar to what I did for perl in the F9 timeframe.
I can do that. Just point me in the right direction please.
Rel-eng should be able to make the tag for you and setup the inheritance. Just open a trac ticket: https://fedorahosted.org/rel-eng/
~spot
On Mon, 2008-11-24 at 13:48 -0600, Les Mikesell wrote:
What perl regression was introduced in F9?
Perl was updated from 5.8 to 5.10 for Fedora 9. If you want to call that a regression, have fun in your little fantasy world.
Jesse Keating wrote:
On Mon, 2008-11-24 at 13:48 -0600, Les Mikesell wrote:
What perl regression was introduced in F9?
Perl was updated from 5.8 to 5.10 for Fedora 9. If you want to call that a regression, have fun in your little fantasy world.
Did that trigger a need to rebuild anything? The only non-backwards-compatible change I've ever seen in perl was when it started to interpolate @ in double-quoted strings between 4.x and 5.x.
On Mon, 2008-11-24 at 14:15 -0600, Les Mikesell wrote:
Perl was updated from 5.8 to 5.10 for Fedora 9. If you want to call that a regression, have fun in your little fantasy world.
Did that trigger a need to rebuild anything? The only non-backwards-compatible change I've ever seen in perl was when it started to interpolate @ in double-quoted strings between 4.x and 5.x.
Yes, for a variety of reasons the packages needed to be rebuilt. Many had to be altered too, as many things were brought into perl core that were addons previously, and rpm spec files had to be adjusted. Spot can fill you in on the more technical details.
On Mon, 2008-11-24 at 12:19 -0800, Jesse Keating wrote:
On Mon, 2008-11-24 at 14:15 -0600, Les Mikesell wrote:
Perl was updated from 5.8 to 5.10 for Fedora 9. If you want to call that a regression, have fun in your little fantasy world.
Did that trigger a need to rebuild anything? The only non-backwards-compatible change I've ever seen in perl was when it started to interpolate @ in double-quoted strings between 4.x and 5.x.
Yes, for a variety of reasons the packages needed to be rebuilt. Many had to be altered too, as many things were brought into perl core that were addons previously, and rpm spec files had to be adjusted. Spot can fill you in on the more technical details.
Perl is extremely major version dependent. Every perl module (basically, anything more complex than a script) had to be rebuilt, and in many cases, upstream bugs had to be filed and fixed before we could retain functionality. I worked aggressively with upstream to both submit the majority of our Fedora only perl patches into the 5.10 tree and to work through the issues discovered during the migration.
End result: We're closer to perl upstream than we've ever been, and we have most of the long-standing perl bugs resolved (and we fixed the "RHEL slow perl" bug without even being aware of it as a byproduct of the methodology).
The fact that you just noticed it means that we must have done some things properly, you're welcome. :)
~spot
On Monday 24 November 2008, Ignacio Vazquez-Abrams wrote:
Within the next few days Python 2.6 will be imported into Rawhide. This means that EVERY single Python-based package in Rawhide will be broken,
Hmm, to clarify: really every one, or "just" every one that installs files into versioned python dirs and/or depends on versioned python(abi)?
If a package installs some *.py, *.pyc, *.pyo somewhere else than in versioned python dirs, and the source *.py is python 2.6 compatible, will the *.pyc and *.pyo compiled with 2.5 break with 2.6? For example, is it necessary/beneficial to rebuild packages like rpmlint (all its python code is either in /usr/bin or /usr/share/rpmlint)?
On Mon, 2008-11-24 at 21:25 +0200, Ville Skyttä wrote:
On Monday 24 November 2008, Ignacio Vazquez-Abrams wrote:
Within the next few days Python 2.6 will be imported into Rawhide. This means that EVERY single Python-based package in Rawhide will be broken,
Hmm, to clarify: really every one, or "just" every one that installs files into versioned python dirs and/or depends on versioned python(abi)?
If a package installs some *.py, *.pyc, *.pyo somewhere else than in versioned python dirs, and the source *.py is python 2.6 compatible, will the *.pyc and *.pyo compiled with 2.5 break with 2.6? For example, is it necessary/beneficial to rebuild packages like rpmlint (all its python code is either in /usr/bin or /usr/share/rpmlint)?
The Python API version (1013) has not changed between 2.5 and 2.6, therefore the bytecode is compatible and .pyc and .pyo files in version-independent locations do not need to be recompiled.
On Mon, 2008-11-24 at 15:04 -0500, Ignacio Vazquez-Abrams wrote:
The Python API version (1013) has not changed between 2.5 and 2.6, therefore the bytecode is compatible and .pyc and .pyo files in version-independent locations do not need to be recompiled.
*sigh*
I just did some actual testing, and I've discovered that this is false. But don't worry, I'll handle it.