FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Redhat > Fedora Development

 
 
LinkBack Thread Tools
 
Old 03-05-2010, 10:12 AM
Michael Schwendt
 
Default Push scripts, mash (was: FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback))

On Fri, 05 Mar 2010 11:03:12 +0100, Kevin wrote:

> Yeah, basically "mash" is a really brute force solution, I think directly
> writing out only the new updates as the first prototypes of Bodhi did and as
> the Extras scripts also did/do is a much smarter solution. Always
> recomputing everything sucks.

It can be beneficial. Also the tagging in koji adds possibilities that are
nice. Such as optional reconstruction of repositories with builds stored
elsewhere.
However, if there are resource constraints, such as slow NFS access
to the packages and repos, rebuilding less sounds more appropriate.

> It was claimed that recomputing is necessary for some obscure multilib
> corner cases.

What might those be?

I could imagine "obsolete multilib packages". They have been pulled in
previously and later become unnecessary for an update and its changed
dep-chain. Old RepoPrune kills them whenever the src.rpm is updated for
the primary arch (=> old multilib pkgs would refer to a src.rpm that
isn't the newest one => prune those). And recomputing the full multilib
dep-chain on demand could still be possible. Deleting packages, which
have been published before, is dangerous, though. Users may have
installed those packages (even if just as automatic dependencies).

> Let me suggest a radical solution for that: drop multilib repos!

Or return to white-listing only a small selection of packages.

There are just too many -devel packages and their dependencies to be ever
relevant to someone for multi-arch installs. Far more users install i686 on
64-bit CPUs, and I have doubts that x86_64 installation users do much
development with i686 packages. At most they install 32-bit apps where
64-bit builds aren't available or "less good".
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 03-05-2010, 04:08 PM
Bill Nottingham
 
Default Push scripts, mash (was: FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback))

Kevin Kofler (kevin.kofler@chello.at) said:
> > So what? That's not twice as much as FE6, which would not have taken
> > several hours to push into such a repo. Not even when running repoclosure
> > on the needsign repo prior to pushing and when updating repoview pages
> > afterwards. Simply because the code that was used worked very differently
> > than "mash".
>
> Yeah, basically "mash" is a really brute force solution, I think directly
> writing out only the new updates as the first prototypes of Bodhi did and as
> the Extras scripts also did/do is a much smarter solution. Always
> recomputing everything sucks.

The issue there is then you have to properly determine what packages
to remove from the repo (unless you just keep everything, which has its
own problems); in this case, recomputing actually makes the code simpler.

> It was claimed that recomputing is necessary for some obscure multilib
> corner cases. Let me suggest a radical solution for that: drop multilib
> repos!

While that would make things simpler and shorter, I doubt it's really
practical. Enough people use and want multilib that I don't think we can
just unilaterally remove it. Moreover, the multilib portion of the compose isn't
the primary time eater.

I certianly don't want to go back to the whitelist case where every time
someone needed a new multilib package we had to update a static whitelist
in the update push tool. That's just silly.

Bill
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 03-05-2010, 04:12 PM
Jesse Keating
 
Default Push scripts, mash (was: FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback))

On Fri, 2010-03-05 at 11:03 +0100, Kevin Kofler wrote:
>
> It was claimed that recomputing is necessary for some obscure multilib
> corner cases. Let me suggest a radical solution for that: drop multilib
> repos! If users really want 32-bit packages, they should enable the 32-bit
> repo. Yes, this will cause file conflicts if you configure yum to always
> drag in both versions by default (exactarch=0). So don't do that. Maybe even
> remove that "feature" from yum altogether. Yum already knows how to fetch
> those 32-bit libs that are actually needed.

Multilib is hard, lets go shopping?


--
Jesse Keating
Fedora -- Freedom˛ is a feature!
identi.ca: http://identi.ca/jkeating
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 03-05-2010, 06:15 PM
Bill Nottingham
 
Default Push scripts, mash (was: FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback))

Kevin Kofler (kevin.kofler@chello.at) said:
> > While that would make things simpler and shorter, I doubt it's really
> > practical. Enough people use and want multilib that I don't think we can
> > just unilaterally remove it. Moreover, the multilib portion of the compose
> > isn't the primary time eater.
> >
> > I certianly don't want to go back to the whitelist case where every time
> > someone needed a new multilib package we had to update a static whitelist
> > in the update push tool. That's just silly.
>
> Why can't we just tell them to add the 32-bit repo to their configuration?
> Possibly even ship fedora-32bit and fedora-updates-32bit configs (disabled
> by default)? With the exactarch=1 setting (the current default), this
> shouldn't be a big problem. The only problem I see is that people would run
> into file conflicts if they use exactarch=0 or yum install
> someapplication.i686, but it's easy to close those as NOTABUG ("sorry,
> multilib is not supported for this package, just use the 64-bit version").
> If those reports become a big problem, isa-based Conflicts tags could be
> added.

Off the top of my head, it would break the install DVD usage case and the wine
usage case. It would also make multilib_policy as a configuration option
meaningless, as you couldn't ever set it to anything other than 'best' and
expect it to work.

Bill
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 03-08-2010, 07:29 PM
Matthew Woehlke
 
Default Push scripts, mash (was: FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback))

Michael Schwendt wrote:
> There are just too many -devel packages and their dependencies to be ever
> relevant to someone for multi-arch installs. Far more users install i686 on
> 64-bit CPUs, and I have doubts that x86_64 installation users do much
> development with i686 packages. At most they install 32-bit apps where
> 64-bit builds aren't available or "less good".

You forget people developing proprietary software... or even just
multilib apps. Multilib is useful if you want to build the 32-bit
version of something on an x86_64 box (and don't want to set up a full
chroot / VM).

(Doubly so for proprietary stuff that may need to build both 32- and
64-bit in the same build tree.)

--
Matthew
Please do not quote my e-mail address unobfuscated in message bodies.
--
Lions and tigers and HIPPOS! Everyone needs a hippo!

--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 03-08-2010, 08:19 PM
Michael Schwendt
 
Default Push scripts, mash (was: FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback))

On Mon, 08 Mar 2010 14:29:42 -0600, Matthew wrote:

> > There are just too many -devel packages and their dependencies to be ever
> > relevant to someone for multi-arch installs. Far more users install i686 on
> > 64-bit CPUs, and I have doubts that x86_64 installation users do much
> > development with i686 packages. At most they install 32-bit apps where
> > 64-bit builds aren't available or "less good".
>
> You forget people developing proprietary software...

Why would development of proprietary software have different requirements
with regard to multilib installations?

When I wrote "...I have doubts that x86_64 installation users do much
development with i686 packages", I didn't exclude developers of
proprietary software either. There may be some who do it actually, but
I don't have any numbers. I only see more users who run into problems
because of the multiarch repos.

> or even just
> multilib apps. Multilib is useful if you want to build the 32-bit
> version of something on an x86_64 box (and don't want to set up a full
> chroot / VM).

The "don't want to" is questionable. Development of the 32-bit version
would still need a full 32-bit test installation. It need not be the
x86_64 box to do full multi-booting instead of VM, but even multi-booting
would be convenient enough, considering how quickly something like Fedora
can be installed. Typical development is not trial-and-error compilation
of both 64-bit and 32-bit and alternating, but rather development on
either arch till something is ready to be built for and to be tested on a
different arch.
Same for multiple target distributions.

> (Doubly so for proprietary stuff that may need to build both 32- and
> 64-bit in the same build tree.)

Again, what special requirements come with the "proprietary" part?
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 03-10-2010, 04:17 PM
Matthew Woehlke
 
Default Push scripts, mash (was: FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback))

Kevin Kofler wrote:
> Matthew Woehlke wrote:
>> You forget people developing proprietary software...
>
> Why would we want to encourage or even support that?

I don't expect Fedora to encourage it (nor should we, IMO)... but that
doesn't change the reality of $DAYJOB. If Fedora drops multilib, I will
be forced to drop Fedora. Not out of spite, but because it will no
longer be able to do my job.

--
Matthew
Please do not quote my e-mail address unobfuscated in message bodies.
--
. . . . . ##@NO CARRIER

--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 03-10-2010, 04:22 PM
Matthew Woehlke
 
Default Push scripts, mash (was: FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback))

Matthew Woehlke wrote:
> Kevin Kofler wrote:
>> Matthew Woehlke wrote:
>>> You forget people developing proprietary software...
>>
>> Why would we want to encourage or even support that?
>
> I don't expect Fedora to encourage it (nor should we, IMO)... but that
> doesn't change the reality of $DAYJOB. If Fedora drops multilib, I will
> be forced to drop Fedora. Not out of spite, but because it will no
> longer be able to do my job.

Meh. Should read "...because it will no longer be possible for me to do
my job with Fedora."

--
Matthew
Please do not quote my e-mail address unobfuscated in message bodies.
--
. . . . . ##@NO CARRIER

--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 03-10-2010, 04:30 PM
Matthew Woehlke
 
Default Push scripts, mash (was: FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback))

Michael Schwendt wrote:
> On Mon, 08 Mar 2010 14:29:42 -0600, Matthew wrote:
>
>>> There are just too many -devel packages and their dependencies to be ever
>>> relevant to someone for multi-arch installs. Far more users install i686 on
>>> 64-bit CPUs, and I have doubts that x86_64 installation users do much
>>> development with i686 packages. At most they install 32-bit apps where
>>> 64-bit builds aren't available or "less good".
>>
>> You forget people developing proprietary software...
>
> Why would development of proprietary software have different requirements
> with regard to multilib installations?

...because said developers are more likely to be developing i686
packages on x86_64.

Mostly, I disagree with your ratio of "people who need multilib" versus
"people for whom multilib causes problems" differently. Probably because
I need multilib and have never experienced multilib-related problems (or
if I have, they were so trivial as to be thoroughly forgettable).

>> Multilib is useful if you want to build the 32-bit version of
>> something on an x86_64 box (and don't want to set up a full chroot
>> / VM).
>
> The "don't want to" is questionable. Development of the 32-bit version
> would still need a full 32-bit test installation.

A test installation of /what you built/, yes. And you have that, since
you just built it. (From that, I guess that you consider testing of a
32-bit program invalid unless done on a pure 32-bit kernel? I sure don't.)

> It need not be the x86_64 box to do full multi-booting instead of
> VM, but even multi-booting would be convenient enough, considering
> how quickly something like Fedora can be installed. Typical
> development is not trial-and-error compilation of both 64-bit and
> 32-bit and alternating, but rather development on either arch till
> something is ready to be built for and to be tested on a different
> arch.

You obviously have a different definition of "typical" than I do.

For $DAYJOB we build both 32- and 64-bit at the same time and test both
within the same test suite. That's my "typical". Given that Windows (go
figure) is the only platform for which we consider 32- versus 64-bit to
be different ports, that's not likely to change.

Multi-booting is not only inconvenient, it isn't an option. Multilib
*is* the method we use to build and test. End of story.

--
Matthew
Please do not quote my e-mail address unobfuscated in message bodies.
--
Sorry, fresh out of .sigs. Maybe tomorrow.
(paraphrased German saying)

--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 03-10-2010, 05:04 PM
Michael Schwendt
 
Default Push scripts, mash (was: FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback))

On Wed, 10 Mar 2010 11:30:05 -0600, Matthew wrote:

> Probably because
> I need multilib and have never experienced multilib-related problems (or
> if I have, they were so trivial as to be thoroughly forgettable).

Just out of interest, does enabling a separate 32-bit repository on a
64-bit installation lead to more/severe problems than using a multiarch
repo?

> (From that, I guess that you consider testing of a
> 32-bit program invalid unless done on a pure 32-bit kernel?

No. I think it depends on what sort of program would be tested.
A 32-bit multlib development environment on a 64-bit installation
does not add a full 32-bit run-time environment.
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 

Thread Tools




All times are GMT. The time now is 12:30 AM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright ©2007 - 2008, www.linux-archive.org