GCC 4.1 Status Report (2005-07-22)

GCC 4.1 Status Report (2005-07-22)

[Mark Mitchell]

We have been in Stage 3 for a little while now.  I'm sure a few more
patches that were proposed in Stage 2 will find their way into 4.1,
but we're approximately feature-complete at this point.  Thank you for
respecting the process.  I'm going to make a call for 4.2 features
when the 4.1 release branch is created, but it's not too early to
start thinking about what you plan to do in 4.2.

There are 225 regressions open against GCC 4.1.  About half of these
(119) are not regressions in 4.0, i.e., they are new regressions
introduced in the course of 4.1.  While it does seem that the
regression rate has declined slightly from 4.0, it still seems rather
high.

As one would expect, given the current focus of the developers, many
of the problems relate to the optimizers: wrong code, crashes, or
failures to optimize.

All in all, I'd say that we look like we're doing OK.  As this is
Stage 3, the focus is now on fixing bugs, and that's certainly what we
need to do.  Stage 3 is scheduled to end September 8th.  I think
that's going to end up slipping, unless we really start knocking down
bugs, but hopefully we can get close.

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304

Re: GCC 4.1 Status Report (2005-07-22)

[Andrew Pinski]

On Jul 22, 2005, at 4:22 PM, Mark Mitchell wrote:
> There are 225 regressions open against GCC 4.1.  About half of these
> (119) are not regressions in 4.0, i.e., they are new regressions
> introduced in the course of 4.1.  While it does seem that the
> regression rate has declined slightly from 4.0, it still seems rather
> high.

I wonder if this includes the ones targeted against 3.4.x which adds
about another 50 or so bugs.

Thanks,
Andrew Pinski

Re: GCC 4.1 Status Report (2005-07-22)

[Mark Mitchell]

Andrew Pinski wrote:
> 
> On Jul 22, 2005, at 4:22 PM, Mark Mitchell wrote:
> 
>> There are 225 regressions open against GCC 4.1.  About half of these
>> (119) are not regressions in 4.0, i.e., they are new regressions
>> introduced in the course of 4.1.  While it does seem that the
>> regression rate has declined slightly from 4.0, it still seems rather
>> high.
> 
> 
> I wonder if this includes the ones targeted against 3.4.x which adds
> about another 50 or so bugs.

You're correct -- I omitted those.  In theory, there should be no such 
bugs; they should all be targeted at 4.0.2 instead, if they apply to 
4.[01].  (That's the policy that Gaby and I have been using.)

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304

Re: GCC 4.1 Status Report (2005-07-22)

[Andrew Pinski]

On Jul 22, 2005, at 5:05 PM, Mark Mitchell wrote:

> Andrew Pinski wrote:
>> On Jul 22, 2005, at 4:22 PM, Mark Mitchell wrote:
>>> There are 225 regressions open against GCC 4.1.  About half of these
>>> (119) are not regressions in 4.0, i.e., they are new regressions
>>> introduced in the course of 4.1.  While it does seem that the
>>> regression rate has declined slightly from 4.0, it still seems rather
>>> high.
>> I wonder if this includes the ones targeted against 3.4.x which adds
>> about another 50 or so bugs.
>
> You're correct -- I omitted those.  In theory, there should be no such 
> bugs; they should all be targeted at 4.0.2 instead, if they apply to 
> 4.[01].  (That's the policy that Gaby and I have been using.)

Should I move them?

Thanks,
Andrew Pinski

Re: GCC 4.1 Status Report (2005-07-22)

[Mark Mitchell]

Andrew Pinski wrote:
> 
> On Jul 22, 2005, at 5:05 PM, Mark Mitchell wrote:
> 
>> Andrew Pinski wrote:
>>
>>> On Jul 22, 2005, at 4:22 PM, Mark Mitchell wrote:
>>>
>>>> There are 225 regressions open against GCC 4.1.  About half of these
>>>> (119) are not regressions in 4.0, i.e., they are new regressions
>>>> introduced in the course of 4.1.  While it does seem that the
>>>> regression rate has declined slightly from 4.0, it still seems rather
>>>> high.
>>>
>>> I wonder if this includes the ones targeted against 3.4.x which adds
>>> about another 50 or so bugs.
>>
>>
>> You're correct -- I omitted those.  In theory, there should be no such 
>> bugs; they should all be targeted at 4.0.2 instead, if they apply to 
>> 4.[01].  (That's the policy that Gaby and I have been using.)
> 
> 
> Should I move them?

Please!  (Otherwise, I'm happy to do it myself.)

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304

Re: GCC 4.1 Status Report (2005-07-22)

[Andrew Pinski]

On Jul 22, 2005, at 5:08 PM, Mark Mitchell wrote:
>
> Please!  (Otherwise, I'm happy to do it myself.)

All done.

-- Pinski

Re: GCC 4.1 Status Report (2005-07-22)

[Mark Mitchell]

Andrew Pinski wrote:
> 
> On Jul 22, 2005, at 5:08 PM, Mark Mitchell wrote:
> 
>>
>> Please!  (Otherwise, I'm happy to do it myself.)
> 
> 
> All done.

Thanks.

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304

Re: GCC 4.1 Status Report (2005-07-22)

[Gerald Pfeifer]

On Fri, 22 Jul 2005, Mark Mitchell wrote:
> We have been in Stage 3 for a little while now.  I'm sure a few more
> patches that were proposed in Stage 2 will find their way into 4.1,
> but we're approximately feature-complete at this point.

I just committed the following update for our main page.  If you'd like
to change "open for bug fixes" to something more strict, please let me
know.

Gerald

Index: index.html
===================================================================
RCS file: /cvs/gcc/wwwdocs/htdocs/index.html,v
retrieving revision 1.511
diff -u -3 -p -r1.511 index.html
--- index.html	25 Jul 2005 10:45:58 -0000	1.511
+++ index.html	25 Jul 2005 11:11:27 -0000
@@ -65,8 +65,8 @@ mission statement</a>.</p>
   will become GCC 4.1.0 (<a href="gcc-4.1/changes.html">current changes</a>)
 </dt><dd>
   Branch status: 
-  <a href="http://gcc.gnu.org/ml/gcc/2005-05/msg00224.html">2005-05-04</a>
-  (<a href="develop.html#stage2">stage 2</a>; open for all maintainers).
+  <a href="http://gcc.gnu.org/ml/gcc/2005-07/msg00954.html">2005-07-22</a>
+  (<a href="develop.html#stage3">stage 3</a>; open for bug fixes).
 </dd>
 
 </dl>

Re: GCC 4.1 Status Report (2005-07-22)

[Mark Mitchell]

Gerald Pfeifer wrote:
> On Fri, 22 Jul 2005, Mark Mitchell wrote:
> 
>>We have been in Stage 3 for a little while now.  I'm sure a few more
>>patches that were proposed in Stage 2 will find their way into 4.1,
>>but we're approximately feature-complete at this point.
> 
> 
> I just committed the following update for our main page.  If you'd like
> to change "open for bug fixes" to something more strict, please let me
> know.

No, I think that's perfect.  Thanks!

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304