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 > Debian > Debian dpkg

 
 
LinkBack Thread Tools
 
Old 02-21-2011, 06:38 AM
Steve Langasek
 
Default x86 triplets and multiarch

On Mon, Feb 21, 2011 at 07:32:19AM +0100, Guillem Jover wrote:

> Given the above we'd need to either switch to i586-linux-gnu or
> i386-linux-gnu, it seems to me both will imply the same amount of
> changes? And thus going for the latter seems the correct solution,
> it matches with the other architectures, can be used as the multiarch
> paths and can reduce some divergence with Ubuntu, all of them a clear
> win!

Why i586 for the multiarch path? That's an arbitrary baseline, based on
what Debian is currently targeting. How do I sell that to the LSB and to
other distributions, most of which AIUI now use i686 as their least common
denominator?

It's not even what Ubuntu is targeting, so you haven't actually reduced any
divergence there:

$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 10.10
Release: 10.10
Codename: maverick
$ dpkg-architecture -ai386 -qDEB_HOST_GNU_TYPE 2>/dev/null
i686-linux-gnu
$

> > Not only for the good, as the switch in Ubuntu to i686 did show,
> > because many configure files assume sse with i686-linux-gnu.

> I'd say any such assumption in those packages is buggy, per above.

Yep, software is buggy. We should be careful not to design a system that
fails because it requires software to not be buggy.

--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
Ubuntu Developer http://www.debian.org/
slangasek@ubuntu.com vorlon@debian.org
 
Old 02-21-2011, 07:43 AM
Guillem Jover
 
Default x86 triplets and multiarch

On Sun, 2011-02-20 at 23:38:36 -0800, Steve Langasek wrote:
> On Mon, Feb 21, 2011 at 07:32:19AM +0100, Guillem Jover wrote:
> > Given the above we'd need to either switch to i586-linux-gnu or
> > i386-linux-gnu, it seems to me both will imply the same amount of
> > changes? And thus going for the latter seems the correct solution,
> > it matches with the other architectures, can be used as the multiarch
> > paths and can reduce some divergence with Ubuntu, all of them a clear
> > win!
>
> Why i586 for the multiarch path? That's an arbitrary baseline, based on
> what Debian is currently targeting. How do I sell that to the LSB and to
> other distributions, most of which AIUI now use i686 as their least common
> denominator?

Hmm probably didn't express myself correctly. I'm proposing to go
for i386-linux-gnu as the GNU triplet for both for Debian and
Ubuntu (well, any derivatives for that matter), which for at least
the paths is also neutral for everyone else (LSB, etc).

I mentioned i586-linux-gnu only because, following Matthias argument,
if we would end up to keep using the cpu baseline in the GNU triplet
for the i386 Debian architecure, it seems we should switch it from
i486-linux-gnu to i586-linux-gnu, so that those applications can
“benefit” from it, in the same way we switched it from i386-linux-gnu
to i486-linux-gnu when the baseline changed long time ago.

So my argument is, given that current applications cannot be currently
assuming i586-linux-gnu anyway, and if we were to consider doing that
change it might take the same amount of effort switching to either
of the triplets, and thus we might as well just switch to what seems
more correct anyway (i386-linux-gnu), which also avoids future
transitions, other reasons on my previous mail, etc.

Hope that makes more sense now?

> > > Not only for the good, as the switch in Ubuntu to i686 did show,
> > > because many configure files assume sse with i686-linux-gnu.
>
> > I'd say any such assumption in those packages is buggy, per above.
>
> Yep, software is buggy. We should be careful not to design a system that
> fails because it requires software to not be buggy.

Well, and buggy software should be fixed, right? Anyway I don't see
how using i386-linux-gnu would make the system fail, at least for the
path part, for the autotools part at most it might make few packages
not choose better optimizations/primitives?

regards,
guillem


--
To UNSUBSCRIBE, email to debian-dpkg-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20110221084340.GA23880@gaara.hadrons.org">http://lists.debian.org/20110221084340.GA23880@gaara.hadrons.org
 
Old 02-21-2011, 04:19 PM
Steve Langasek
 
Default x86 triplets and multiarch

On Mon, Feb 21, 2011 at 09:43:41AM +0100, Guillem Jover wrote:
> On Sun, 2011-02-20 at 23:38:36 -0800, Steve Langasek wrote:
> > On Mon, Feb 21, 2011 at 07:32:19AM +0100, Guillem Jover wrote:
> > > Given the above we'd need to either switch to i586-linux-gnu or
> > > i386-linux-gnu, it seems to me both will imply the same amount of
> > > changes? And thus going for the latter seems the correct solution,
> > > it matches with the other architectures, can be used as the multiarch
> > > paths and can reduce some divergence with Ubuntu, all of them a clear
> > > win!

> > Why i586 for the multiarch path? That's an arbitrary baseline, based on
> > what Debian is currently targeting. How do I sell that to the LSB and to
> > other distributions, most of which AIUI now use i686 as their least common
> > denominator?

> Hmm probably didn't express myself correctly. I'm proposing to go
> for i386-linux-gnu as the GNU triplet for both for Debian and
> Ubuntu (well, any derivatives for that matter), which for at least
> the paths is also neutral for everyone else (LSB, etc).

Oops - reading failure on my part, sorry.

But then I guess the converse problem is, what software makes assumptions
about the difference between i386-linux-gnu and i486-linux-gnu that will
hurt us?

> > > I'd say any such assumption in those packages is buggy, per above.

> > Yep, software is buggy. We should be careful not to design a system that
> > fails because it requires software to not be buggy.

> Well, and buggy software should be fixed, right?

Yes, but it makes me nervous that fixing such software might be on the
critical path for multiarch.

> Anyway I don't see how using i386-linux-gnu would make the system fail, at
> least for the path part, for the autotools part at most it might make few
> packages not choose better optimizations/primitives?

Maybe, but considering some of the features lacking in i386, that could be a
pretty severe regression, causing the system to fail the "fit for purpose"
test.

--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
Ubuntu Developer http://www.debian.org/
slangasek@ubuntu.com vorlon@debian.org
 

Thread Tools




All times are GMT. The time now is 05:19 AM.

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