Re: What are the chances of another release from gcc-3_2-branch?

Re: What are the chances of another release from gcc-3_2-branch?

[John David Anglin]

> If we are going to do it, I would suggest freezing what we have (unless
> there's a very good argument for one or two more fixes), do a prerelease
> tarball, get some testing of that and ship.

I have a fix for parisc-linux that I would like to backport from 3.3.
It's not a regression, but 3.3 or later won't bootstrap with 3.2 because
of continued code growth in expr.c.  The change is small and the impact
is limited to the pa.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

Re: What are the chances of another release from gcc-3_2-branch?

[Michael Elizabeth Chastain]

For what it's worth, my gdb testing shows no regressions between
gcc 3.2 and gcc gcc-3_2-branch%20021224 on native i686-pc-linux-gnu.

That is the *gdb* test suite; I'm not running gcc test suites yet.

Michael C

Re: What are the chances of another release from gcc-3_2-branch?

[Volker Reichelt]

I'm all in favor for a 3.2.2 release, but I would suggest to wait with a
release until more bugs get fixed.

Currently we have about 34 bugs in GNATS that are marked as 3.2/3.3
regressions. Since (hopefully) many of them will be fixed for 3.3 anyway,
it would be nice to have the patches backported for the 3.2 branch
(at least the easy ones).

Even releasing 3.2.2 a couple of weeks after 3.3 in order not to have
two releases in the same timeframe would be fine IMHO.

Regards,
Volker

Re: What are the chances of another release from gcc-3_2-branch?

[Michael Elizabeth Chastain]

Mark Mitchell writes:
> I don't think any decision has been made.  Personally, I'm trying to focus
> on 3.3/3.4; there's plenty enough to do there.  But, if people want another
> 3.2 release we can do one.

Thanks, this helps me plan.

Michael C

Re: What are the chances of another release from gcc-3_2-branch?

[Kaveh R. Ghazi]

 > I don't think any decision has been made.  Personally, I'm trying to
 > focus on 3.3/3.4; there's plenty enough to do there.  But, if people
 > want another 3.2 release we can do one.

Here's what turned up with a quick grep in the various ChangeLogs
since 3.2.1.  We've got around 30 PRs fixed plus a bunch of other
non-PR related fixes.

I think the decision of whether to spin another release depends on how
serious are the problems these fixes address.  (I haven't looked myself.)

		--Kaveh

gcc:
        PR opt/8794
        PR other/8882
        PR c/7622
        PR preprocessor/8524
        PR 8362
        PR c/8639
        PR optimization/8275
        PR c/8588
        PR c/8518
        PR c/8439
        PR optimization/8599

cp:
        PR c++/8442
        PR C++/8031
        PR c++/8372
        PR C++/8799
        PR c++/8615
        PR c++/5919
        PR c++/8727
        PR c++/8663
        PR c++/8332
        PR c++/8493
        PR c++/8214

libstdc++-v3:
        PR libstdc++/8887
        PR libstdc++/8230
        PR libstdc++/8708
        PR libstdc++/8790
        PR libstdc++/7445
        PR libstdc++/6745
	PR libstdc++/8399
        PR libstdc++/8399
        PR libstdc++/8230
--
Kaveh R. Ghazi			ghazi@caip.rutgers.edu

Re: What are the chances of another release from gcc-3_2-branch?

[Tom Tromey]

>>>>> "Kaveh" == Kaveh R Ghazi <ghazi@caip.rutgers.edu> writes:

Kaveh> Here's what turned up with a quick grep in the various ChangeLogs
Kaveh> since 3.2.1.  We've got around 30 PRs fixed plus a bunch of other
Kaveh> non-PR related fixes.

It occurs to me that we could enhance the cvs commit script to record
the branch on which a PR-related checkin is made.  Then we could do a
gnats query to pick up this info.  What do you think of this idea?

Tom

Re: What are the chances of another release from gcc-3_2-branch?

[Kaveh R. Ghazi]

 > From: Tom Tromey <tromey@redhat.com>
 > 
 > >>>>> "Kaveh" == Kaveh R Ghazi <ghazi@caip.rutgers.edu> writes:
 > 
 > Kaveh> Here's what turned up with a quick grep in the various ChangeLogs
 > Kaveh> since 3.2.1.  We've got around 30 PRs fixed plus a bunch of other
 > Kaveh> non-PR related fixes.
 > 
 > It occurs to me that we could enhance the cvs commit script to record
 > the branch on which a PR-related checkin is made.  Then we could do a
 > gnats query to pick up this info.  What do you think of this idea?
 > Tom

Sounds like it would make generating the release notes easier for
Joe. ;-)

--
Kaveh R. Ghazi			ghazi@caip.rutgers.edu

What are the chances of another release from gcc-3_2-branch?

[Michael Elizabeth Chastain]

(I asked this on gcc-help first and was directed here.)

I run a gdb test bed with several versions of gcc.  I note that
gcc-3_3-branch has been created, so I will pick it up and
incorporate it in my test bed.

My question: is it safe for me to stop testing gcc-3_2-branch?  That is,
what are the chances of another gcc release such as a "gcc 3.2.2" coming
from the gcc-3_2-branch?

I note that several of the ChangeLog's have been updated since
the release of gcc 3.2.1 on 2002-11-20:

  gcc/f/ChangeLog
  gcc/cp/ChangeLog
  libstdc++-v3/ChangeLog
  gcc/ChangeLog
  gcc/testsuite/ChangeLog

I'm just looking for some insight into the branch status here.
My default plan is to keep testing with gcc-3_2-branch even after
gcc-3_3-branch is created, as it appears to be active.

Thanks,

Michael C

Re: What are the chances of another release from gcc-3_2-branch?

[Phil Edwards]

On Sun, Dec 15, 2002 at 10:42:01PM -0600, Michael Elizabeth Chastain wrote:
> My question: is it safe for me to stop testing gcc-3_2-branch?  That is,
> what are the chances of another gcc release such as a "gcc 3.2.2" coming
> from the gcc-3_2-branch?
> 
> I note that several of the ChangeLog's have been updated since
> the release of gcc 3.2.1 on 2002-11-20:
> 
>   gcc/f/ChangeLog
>   gcc/cp/ChangeLog
>   libstdc++-v3/ChangeLog
>   gcc/ChangeLog
>   gcc/testsuite/ChangeLog
> 
> I'm just looking for some insight into the branch status here.
> My default plan is to keep testing with gcc-3_2-branch even after
> gcc-3_3-branch is created, as it appears to be active.

Likewise.  I'd like to see at least one more release of 3.2.  A number of
bugs have been fixed on the branch since 3.2.1.


Phil

-- 
I would therefore like to posit that computing's central challenge, viz. "How
not to make a mess of it," has /not/ been met.
                                                 - Edsger Dijkstra, 1930-2002

Re: What are the chances of another release from gcc-3_2-branch?

[Mark Mitchell]

--On Thursday, December 19, 2002 12:54:13 AM -0500 Phil Edwards 
<phil@jaj.com> wrote:

> On Sun, Dec 15, 2002 at 10:42:01PM -0600, Michael Elizabeth Chastain
> wrote:
>> My question: is it safe for me to stop testing gcc-3_2-branch?  That is,
>> what are the chances of another gcc release such as a "gcc 3.2.2" coming
>> from the gcc-3_2-branch?

I don't think any decision has been made.  Personally, I'm trying to focus
on 3.3/3.4; there's plenty enough to do there.  But, if people want another
3.2 release we can do one.

If someone else wants to step forward to run a 3.2.2 release, that would
be fine, too.

Thanks,

-- 
Mark Mitchell                mark@codesourcery.com
CodeSourcery, LLC            http://www.codesourcery.com

Re: What are the chances of another release from gcc-3_2-branch?

[Joe Buck]

On Sun, Dec 15, 2002 at 10:42:01PM -0600, Michael Elizabeth Chastain
wrote:
> >> My question: is it safe for me to stop testing gcc-3_2-branch?  That is,
> >> what are the chances of another gcc release such as a "gcc 3.2.2" coming
> >> from the gcc-3_2-branch?

Mark Mitchell writes:

> I don't think any decision has been made.  Personally, I'm trying to focus
> on 3.3/3.4; there's plenty enough to do there.  But, if people want another
> 3.2 release we can do one.
> 
> If someone else wants to step forward to run a 3.2.2 release, that would
> be fine, too.

I'd started a release-notes document on the assumption that there would be
a 3.2.2.  That doesn't imply that therefore we have to do one :-), but it
does mean that I have some data.

Currently it appears that there are 24 PR fixes that have been added to
the 3.2 branch since the 3.2.1 release; 14 of these are ICEs, and 9 are
C++ compiler or library bug fixes.  It seems that *someone* expects
something to happen from all this work.  Distros relying on 3.2 compilers
might want to pick up these fixes.

Some of the fixes are for obscure cases, but some of them people are
likely to want to pick up.  Here are some highlights:

ICEs:
8332	(c++) builtin strlen/template interation causes ICE
8439	(c, not c++) empty struct causes ICE
8518	ICE when compiling mplayer ("extern inline" issue)
8615	(c++) ICE with out-of-range character constant template argument

C++/libstdc++:
8214	conversion from const char* const to char* sometimes accepted illegally
7445	poor performance of std::locale::classic() in multi-threaded applications
8399	sync_with_stdio(false) breaks unformatted input
8790	Use of non-thread-safe strtok in src/localename.cc
8887	Bug in date formats with --enable-clocale=generic

Optimization:
8794	optimization improperly eliminates certain expressions

Preprocessor:
8524	_Pragma within macros is improperly expanded

x86-specific (Intel/AMD):

8588	ICE in extract_insn, at recog.c:NNNN (shift instruction)

Ideally Mark shouldn't be asked to do a 3.2.2 and 3.3 in about the same
timeframe.  If someone else wants to volunteer, that would be great.
If we are going to do it, I would suggest freezing what we have (unless
there's a very good argument for one or two more fixes), do a prerelease
tarball, get some testing of that and ship.

Re: What are the chances of another release from gcc-3_2-branch?

[Eric Botcazou]

> If we are going to do it, I would suggest freezing what we have (unless
> there's a very good argument for one or two more fixes), do a prerelease
> tarball, get some testing of that and ship.

I have one: we still have regressions from the 3.2 release regarding loops 
compiled with -O2 -funroll-loops on K6 (PR opt/8599) and powerpc. The patch 
is awaiting review and testing on powerpc here:
http://gcc.gnu.org/ml/gcc-patches/2002-12/msg00787.html
http://gcc.gnu.org/ml/gcc-patches/2002-12/msg00875.html

-- 
Eric Botcazou

Re: What are the chances of another release from gcc-3_2-branch?

[Gabriel Dos Reis]

Mark Mitchell <mark@codesourcery.com> writes:

| --On Thursday, December 19, 2002 12:54:13 AM -0500 Phil Edwards
| <phil@jaj.com> wrote:
| 
| > On Sun, Dec 15, 2002 at 10:42:01PM -0600, Michael Elizabeth Chastain
| > wrote:
| >> My question: is it safe for me to stop testing gcc-3_2-branch?  That is,
| >> what are the chances of another gcc release such as a "gcc 3.2.2" coming
| >> from the gcc-3_2-branch?
| 
| I don't think any decision has been made.  Personally, I'm trying to focus
| on 3.3/3.4; there's plenty enough to do there.  But, if people want another
| 3.2 release we can do one.
| 
| If someone else wants to step forward to run a 3.2.2 release, that would
| be fine, too.

I think a 3.2.2 release will be of great help as it has non-negligible
fixes.

If the only obstacle for having such a release is lack of time, then I
would volunteer for making that release provided I'm explained the
necessary steps.

-- Gaby

Re: What are the chances of another release from gcc-3_2-branch?

[Gerald Pfeifer]

On Sun, 15 Dec 2002, Michael Elizabeth Chastain wrote:
> My question: is it safe for me to stop testing gcc-3_2-branch?  That is,
> what are the chances of another gcc release such as a "gcc 3.2.2" coming
> from the gcc-3_2-branch?

High, I'd say, but this usually depends on the feedback users and
developers give to Mark (and the community), and also on how long
it'll take to stabilize the 3.3 branch.

> I'm just looking for some insight into the branch status here.
> My default plan is to keep testing with gcc-3_2-branch even after
> gcc-3_3-branch is created, as it appears to be active.

That sounds like a good plan, especially in the (hopefully unlikely)
case we add regressions on the 3.2 branch.

Gerald
-- 
Gerald "Jerry"   pfeifer@dbai.tuwien.ac.at   http://www.pfeifer.com/gerald/