GCC-3.2.3 prerelease, 1st iteration

GCC-3.2.3 prerelease, 1st iteration

[Gabriel Dos Reis]

Hi,

  I uploaded GCC-3.2.3 prerelease tarballs on the FTP server as

  ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*3.2.3-20030412.tar.*

Please donwload and test them. Report problems to me and fill GNATS
bug-reports.

(Internet connectivity was not as expected the last couple of weeks and
I could not *reliably* make prereleases tarballs.  Sorry about that).


-- Gaby

Re: GCC-3.2.3 prerelease, 1st iteration

[Eric Botcazou]

> Please donwload and test them. Report problems to me and fill GNATS
> bug-reports.

Bootstrap broken on sparc-sun-solaris2.5.1 (regression from GCC 3.2.2) and 
install broken on multilib sparc-sun-solaris2.[7-9] for some systems (not a 
regression from GCC 3.2.2).

I should quickly have solutions for both.

-- 
Eric Botcazou

Re: GCC-3.2.3 prerelease, 1st iteration

[Christian Joensson]

On Sat, Apr 12, 2003 at 08:37:46PM +0200, Gabriel Dos Reis wrote:
> 
> Hi,
> 
>   I uploaded GCC-3.2.3 prerelease tarballs on the FTP server as
> 
>   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*3.2.3-20030412.tar.*
> 

Red Hat Linux 9 (Shrike) ia32 dual pentium III i686:

binutils-2.13.90.0.18-9
bison-1.35-6
dejagnu-1.4.2-10
expect-5.38.0-88
gcc-3.2.2-5
glibc-2.3.2-27.9
glibc-kernheaders-2.4-8.10
kernel-smp-2.4.20-9
tcl-8.3.5-88

configure
flags: --host=i686-linux --prefix=/usr/local/gcc-binutils
--enable-__cxa_atexit --enable-shared --enable-threads=posix
--enable-languages=c,c++,f77,java,objc

Tests posted at
http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg00739.html and witf
FAIL in

FAIL: objc.dg/naming-1.m  (test for errors, line 20)
FAIL: objc.dg/naming-1.m (test for excess errors)
FAIL: objc.dg/naming-2.m  (test for errors, line 7)
FAIL: objc.dg/naming-2.m (test for excess errors)

FAIL: 26_numerics/c99_classification_macros_c.cc (test for excess errors)
WARNING: program timed out.
FAIL: thread/pthread4.cc execution test

Cheers,

/ChJ

Re: GCC-3.2.3 prerelease, 1st iteration

[Christian Joensson]

On Sat, Apr 12, 2003 at 08:37:46PM +0200, Gabriel Dos Reis wrote:
> 
> Hi,
> 
>   I uploaded GCC-3.2.3 prerelease tarballs on the FTP server as
> 
>   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*3.2.3-20030412.tar.*
> 

This was on a Windows XP SP1/cygwin P4 system with these packages:

binutils  2.13.90 20030412
bison	  1.75-1
cygwin    1.3.21-1
dejagnu   1.4.3       (cygwin ver. 20010117-1 with 1.4.3 on top)
gcc       3.4 20030412 (experimental)

configure flags: --host=i686-pc-cygwin
--prefix=/usr/local/gcc-binutils --enable-shared
--enable-threads=posix

Tests posted at
http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg00739.html and with
FAIL in

FAIL: g++.dg/abi/rtti1.C scan-assembler-dem-not \\ntypeinfo for A[: \\t\\n]
FAIL: g++.dg/ext/instantiate3.C scan-assembler-not 
FAIL: g++.dg/special/initp1.C execution test
FAIL: g++.eh/badalloc1.C  Execution test
FAIL: g++.law/profile1.C (test for excess errors)

FAIL: gcc.dg/20021014-1.c (test for excess errors)
WARNING: gcc.dg/20021014-1.c compilation failed to produce executable
FAIL: gcc.dg/typeof-2.c (test for excess errors)
FAIL: gcc.dg/weak-9.c (test for excess errors)

FAIL: objc.dg/naming-1.m  (test for errors, line 20)
FAIL: objc.dg/naming-1.m (test for excess errors)
FAIL: objc.dg/naming-2.m  (test for errors, line 7)
FAIL: objc.dg/naming-2.m (test for excess errors)

FAIL: 26_numerics/c99_classification_macros_c.cc (test for excess errors)
FAIL: 27_io/filebuf_members.cc execution test
FAIL: 27_io/istream_extractor_arith/12.cc execution test

Some of these FAILures may very well be due to problems with
binutils... see
http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg00724.html for the
installed binutils's testresults together witg the installed gcc's...

Cheers,

/ChJ

Re: GCC-3.2.3 prerelease, 1st iteration

[Christian Joensson]

On Sat, Apr 12, 2003 at 08:37:46PM +0200, Gabriel Dos Reis wrote:
> 
> Hi,
> 
>   I uploaded GCC-3.2.3 prerelease tarballs on the FTP server as
> 
>   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*3.2.3-20030412.tar.*
> 

Aurora SPARC Linux 1.0 (Ansel) Ultra1 single UltraSparc I sun4u:

binutils 2.13.90 20030411 sparc64-linux
bison-1.35-1
dejagnu-1.4.2-10 (ftp://ftp.redhat.com/pub/redhat/linux/rawhide/SRPMS/SRPMS)
expect-5.32.2-67
gcc 3.3 20030411 (prerelease)
glibc-2.2.5-42
glibc-kernheaders-2.4-7.16sparc
glibc64-2.2.5-42
kernel-2.4.20-3.0sparc (experimental Aurora kernel)
tcl-8.3.3-67

configure flags: --host=sparc64-linux --prefix=/usr/local/gcc-binutils
--enable-shared --enable-threads=posix
--enable-languages=c,c++,f77,java,objc

Tests posted at
http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg00781.html
FAIL in

FAIL: gcc.dg/20021014-1.c execution test
FAIL: linkage.c

FAIL: g++.law/profile1.C  Execution test

FAIL: objc.dg/naming-1.m  (test for errors, line 20)
FAIL: objc.dg/naming-1.m (test for excess errors)
FAIL: objc.dg/naming-2.m  (test for errors, line 7)
FAIL: objc.dg/naming-2.m (test for excess errors)


FAIL: 26_numerics/c99_classification_macros_c.cc (test for excess errors)
WARNING: program timed out.
FAIL: thread/pthread4.cc execution test


Cheers,

/ChJ

Re: GCC-3.2.3 prerelease, 1st iteration

[Gerald Pfeifer]

On Sat, 12 Apr 2003, Eric Botcazou wrote:
>> Please donwload and test them. Report problems to me and fill GNATS
>> bug-reports.
> Bootstrap broken on sparc-sun-solaris2.5.1 (regression from GCC 3.2.2)

Is your bootstrap failure the same as what I'm seeing on
sparc-sun-solaris2.8?

 c-parse.y: At top level:
 c-parse.y:6175: redefinition of `make_pointer_declarator'
 c-parse.y:3084: `make_pointer_declarator' previously defined here

> I should quickly have solutions for both.

Great, thanks!

Gerald

Re: GCC-3.2.3 prerelease, 1st iteration

[Eric Botcazou]

> Is your bootstrap failure the same as what I'm seeing on
> sparc-sun-solaris2.8?
>
>  c-parse.y: At top level:
>  c-parse.y:6175: redefinition of `make_pointer_declarator'
>  c-parse.y:3084: `make_pointer_declarator' previously defined here

No, mine is a libiberty configure problem, which is specific to Solaris 2.5.1 
because of an outdated header file. Kaveh posted a patch to solve it.

Note that I've successfully bootstrapped the 3.2 branch from CVS (not the 
pre-release tarball) + Kaveh's patch this morning on Solaris 2.5.1, 2.6, 2.7 
and 2.8 without a hitch.

-- 
Eric Botcazou

Re: GCC-3.2.3 prerelease, 1st iteration

[Gabriel Dos Reis]

Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at> writes:

| On Sat, 12 Apr 2003, Eric Botcazou wrote:
| >> Please donwload and test them. Report problems to me and fill GNATS
| >> bug-reports.
| > Bootstrap broken on sparc-sun-solaris2.5.1 (regression from GCC 3.2.2)
| 
| Is your bootstrap failure the same as what I'm seeing on
| sparc-sun-solaris2.8?

Gerald, please, could you send me a report + filing GNATS?

Many thanks,

-- Gaby

Re: GCC-3.2.3 prerelease, 1st iteration

[Gabriel Dos Reis]

Christian Joensson <c.christian.joensson@telia.com> writes:

[...]

| Tests posted at
| http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg00739.html and witf
| FAIL in
| 
| FAIL: objc.dg/naming-1.m  (test for errors, line 20)
| FAIL: objc.dg/naming-1.m (test for excess errors)
| FAIL: objc.dg/naming-2.m  (test for errors, line 7)
| FAIL: objc.dg/naming-2.m (test for excess errors)

These seems to be caused by unsifficient pruning of diagnostics.  Will
be fixed. 

| FAIL: 26_numerics/c99_classification_macros_c.cc (test for excess errors)
| WARNING: program timed out.

This is random failure depending on the target's setting. 

| FAIL: thread/pthread4.cc execution test

This is new to me.

Thanks,

-- Gaby

Re: GCC-3.2.3 prerelease, 1st iteration

[Gerald Pfeifer]

On Mon, 14 Apr 2003, Eric Botcazou wrote:
>>  c-parse.y: At top level:
>>  c-parse.y:6175: redefinition of `make_pointer_declarator'
>>  c-parse.y:3084: `make_pointer_declarator' previously defined here
> Note that I've successfully bootstrapped the 3.2 branch from CVS (not
> the pre-release tarball) + Kaveh's patch this morning on Solaris 2.5.1,
> 2.6, 2.7 and 2.8 without a hitch.

It seems we have some parallel build issue (or it's just because my
srcdir is NFS-mounted, while objdir is local), at least I could now
bootstrap after manually removing c-parse.y and c-parse.in.

As also my experience was using CVS...

On Mon, 14 Apr 2003, Gabriel Dos Reis wrote:
> Gerald, please, could you send me a report + filing GNATS?

...Gaby, the release itself (using tar-balls) shouldn't be affected,
and it seems the problem is even hard to trigger (when I tried now,
several times).

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

Re: GCC-3.2.3 prerelease, 1st iteration

[Joe Buck]

On Sat, Apr 12, 2003 at 08:37:46PM +0200, Gabriel Dos Reis wrote:
>   I uploaded GCC-3.2.3 prerelease tarballs on the FTP server as
> 
>   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*3.2.3-20030412.tar.*
> 
> Please donwload and test them. Report problems to me and fill GNATS
> bug-reports.

Successfully built and tested on a Red Hat 8.0 box.  All languages
including Ada were built, though I don't have the Ada test suite.
Results and system details are at

http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg00825.html

Note: I first tried "make -j2 bootstrap-lean", but got failures in
the Ada build that went away when I avoided -j2.  I've lost the log
with the failure and the box is at home, I could try it again if
this is not a known bug. (I regularly use -jn when doing builds that
do not include Ada).

Re: GCC-3.2.3 prerelease, 1st iteration

[Eric Botcazou]

> | FAIL: 26_numerics/c99_classification_macros_c.cc (test for excess
> | errors) WARNING: program timed out.
>
> This is random failure depending on the target's setting.
>
> | FAIL: thread/pthread4.cc execution test
>
> This is new to me.

I think you didn't split correctly, it's

WARNING: program timed out.
FAIL: thread/pthread4.cc execution test

I have seen this "problem" since I started testing on Sparc boxes (Ultra-2 
and Ultra-60).

-- 
Eric Botcazou

Re: GCC-3.2.3 prerelease, 1st iteration

[Loren James Rittle]

In article <200304151157.55166.ebotcazou@libertysurf.fr> you write:

> I think you didn't split correctly, it's

> WARNING: program timed out.
> FAIL: thread/pthread4.cc execution test

> I have seen this "problem" since I started testing on Sparc boxes (Ultra-2 
> and Ultra-60).

Yes, as the author/installer of the test: It is probably harmless if
the test fails with the "program timed out" message before the FAIL.
It would be useful to check to see if it is "live/deadlocking", or
really just not completing all iterations within the dejagnu time limit.

Regards,
Loren

Re: GCC-3.2.3 prerelease, 1st iteration

[Christian Joensson]

On Tue, Apr 15, 2003 at 05:23:54AM -0500, Loren James Rittle wrote:
> In article <200304151157.55166.ebotcazou@libertysurf.fr> you write:
> 
> > I think you didn't split correctly, it's
> 
> > WARNING: program timed out.
> > FAIL: thread/pthread4.cc execution test
> 
> > I have seen this "problem" since I started testing on Sparc boxes (Ultra-2 
> > and Ultra-60).
> 
> Yes, as the author/installer of the test: It is probably harmless if
> the test fails with the "program timed out" message before the FAIL.
> It would be useful to check to see if it is "live/deadlocking", or
> really just not completing all iterations within the dejagnu time limit.

I use 

set_board_info gcc,timeout 1200
set_board_info sim,timeout 1200

in /usr/share/dejagnu/baseboards/unix.exp to fix timeouts in the gcc
testsuite but I don't know how to set any such variable to get the
libstdc++-v3 testsuite use different time-out value than the 300
default one.. any one?


Cheers,

/ChJ

Re: GCC-3.2.3 prerelease, 2nd iteration

[Gabriel Dos Reis]

  I uploaded a second version of GCC-3.2.3 prerelease tarballs on the
FTP server as 

  ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*3.2.3-20030416.tar.*

Please donwload and test them. Report problems to me and fill GNATS
bug-reports.

This pre-release is likely to be the last one before the actual 3.2.3
release.  It includes a patch to prune "syntax error" for the
Objective-C testsuite.  The bootstrap issue on MingW is still open --
I got no analyzis/patch and right now I don't have access to such an
environment. Any help is welcome.

Thanks,

-- Gaby

Re: GCC-3.2.3 prerelease, 2nd iteration

[Christian Joensson]

On Wed, Apr 16, 2003 at 10:51:42AM +0200, Gabriel Dos Reis wrote:
> 
>   I uploaded a second version of GCC-3.2.3 prerelease tarballs on the
> FTP server as 
> 
>   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*3.2.3-20030416.tar.*
> 
> Please donwload and test them. Report problems to me and fill GNATS
> bug-reports.
> 

Test results for i686-linux posted at
http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg00994.html

These are the only tests that unexpectedly fail...

FAIL: 26_numerics/c99_classification_macros_c.cc (test for excess errors)
WARNING: program timed out.
FAIL: thread/pthread4.cc execution test

Cheers,

/ChJ

Re: GCC-3.2.3 prerelease, 2nd iteration

[Christian Joensson]

On Wed, Apr 16, 2003 at 10:51:42AM +0200, Gabriel Dos Reis wrote:
> 
>   I uploaded a second version of GCC-3.2.3 prerelease tarballs on the
> FTP server as 
> 
>   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*3.2.3-20030416.tar.*
> 
> Please donwload and test them. Report problems to me and fill GNATS
> bug-reports.

Test results for i686-cygwin posted at
http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg01002.html

These are the tests that unexpectedly fail...


FAIL: g++.dg/abi/rtti1.C scan-assembler-dem-not \\ntypeinfo for A[: \\t\\n]
FAIL: g++.dg/ext/instantiate3.C scan-assembler-not
FAIL: g++.dg/special/initp1.C execution test
FAIL: g++.eh/badalloc1.C  Execution test

FAIL: g++.law/profile1.C (test for excess errors)

FAIL: gcc.dg/20021014-1.c (test for excess errors)
WARNING: gcc.dg/20021014-1.c compilation failed to produce executable
FAIL: gcc.dg/typeof-2.c (test for excess errors)
FAIL: gcc.dg/weak-9.c (test for excess errors)

FAIL: 26_numerics/c99_classification_macros_c.cc (test for excess errors)
FAIL: 27_io/filebuf_members.cc execution test
FAIL: 27_io/istream_extractor_arith/12.cc execution test

The failures might very well be related to failuers in the binutils
tools...

Cheers,

/ChJ

Re: GCC-3.2.3 prerelease, 2nd iteration

[Kaveh R. Ghazi]

 > I uploaded a second version of GCC-3.2.3 prerelease tarballs on the
 > FTP server as
 > 
 >   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*3.2.3-20030416.tar.*
 > 
 > Please donwload and test them. Report problems to me and fill GNATS
 > bug-reports.

I successfully tested gcc 3.2.3 20030416 (prerelease) on:

mips-sgi-irix6.5	http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg01151.html
mips-sgi-irix6.2	http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg01152.html
sparc-sun-solaris2.7	http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg01153.html
sparc64-sun-solaris2.7	http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg01154.html

None of the failures appear to be regressions from any previous 3.2.x.

All's clear from my standpoint. :-)

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

Re: GCC-3.2.3 prerelease, 1st iteration

[Gabriel Dos Reis]

Gabriel Dos Reis <gdr@integrable-solutions.net> writes:

| Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at> writes:
| 
| | On Mon, 14 Apr 2003, Jakub Jelinek wrote:
| | > Those are likely because you've upgraded bison.
| | >
| | > 2003-03-17  Zack Weinberg  <zack@codesourcery.com>
| | >
| | >         * objc.dg/naming-1.m: Use "(parse|syntax) error".
| | >         * objc.dg/naming-2.m: Likewise.
| | >
| | > should fix it, maybe it should be applied to gcc-3_2-branch too.
| | 
| | Yes, that seems to be necessary.  Zack,would you mind doing that?
| 
| Done with the below:

Actually not.  I got the following:

  Cannot open /sourceware/cvs-tmp/#gcc.lastdir.31503: Permission denied
  cvs server: Pre-commit check failed
  Cannot open /sourceware/cvs-tmp/#gcc.lastdir.31503: Permission denied
  cvs server: Pre-commit check failed

Does anyone have a clue on what is going on on gcc.gnu.org side?


Thanks,

-- Gaby

Re: GCC-3.2.3 prerelease, 1st iteration

[Gabriel Dos Reis]

Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at> writes:

| On Mon, 14 Apr 2003, Jakub Jelinek wrote:
| > Those are likely because you've upgraded bison.
| >
| > 2003-03-17  Zack Weinberg  <zack@codesourcery.com>
| >
| >         * objc.dg/naming-1.m: Use "(parse|syntax) error".
| >         * objc.dg/naming-2.m: Likewise.
| >
| > should fix it, maybe it should be applied to gcc-3_2-branch too.
| 
| Yes, that seems to be necessary.  Zack,would you mind doing that?

Done with the below:

-- Gaby
Index: ChangeLog
===================================================================
RCS file: /cvs/gcc/gcc/gcc/testsuite/ChangeLog,v
retrieving revision 1.1672.2.166.2.124
diff -p -r1.1672.2.166.2.124 ChangeLog
*** ChangeLog	30 Mar 2003 04:45:06 -0000	1.1672.2.166.2.124
--- ChangeLog	16 Apr 2003 02:11:35 -0000
***************
*** 1,3 ****
--- 1,11 ----
+ 2003-04-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
+         Backport from mainline.
+ 
+         2003-03-17  Zack Weinberg  <zack@codesourcery.com>
+ 
+ 	        * objc.dg/naming-1.m: Use "(parse|syntax) error".
+ 	        * objc.dg/naming-2.m: Likewise.
+ 
  2003-03-29  Kaveh R. Ghazi  <ghazi@caip.rutgers.edu>
  
  	* gcc.c-torture/execute/20020307-2.x: New file, XFAIL mips-irix6.
Index: objc.dg/naming-1.m
===================================================================
RCS file: /cvs/gcc/gcc/gcc/testsuite/objc.dg/naming-1.m,v
retrieving revision 1.1
diff -p -r1.1 naming-1.m
*** objc.dg/naming-1.m	1 Aug 2001 08:15:11 -0000	1.1
--- objc.dg/naming-1.m	16 Apr 2003 02:11:37 -0000
*************** void foo(void)
*** 17,21 ****
  void bar(void)
  {
  	View *view;	/* ok */
! 	View = 1;	/* { dg-error "parse error" } */
  }
--- 17,21 ----
  void bar(void)
  {
  	View *view;	/* ok */
! 	View = 1;	/* { dg-error "(parse|syntax) error" } */
  }
Index: objc.dg/naming-2.m
===================================================================
RCS file: /cvs/gcc/gcc/gcc/testsuite/objc.dg/naming-2.m,v
retrieving revision 1.1
diff -p -r1.1 naming-2.m
*** objc.dg/naming-2.m	1 Aug 2001 08:15:11 -0000	1.1
--- objc.dg/naming-2.m	16 Apr 2003 02:11:37 -0000
***************
*** 4,10 ****
  
  @interface Foo
  @end
! float Foo;  /* { dg-error "parse error before" } */
  
  double Bar;
  @interface Bar
--- 4,10 ----
  
  @interface Foo
  @end
! float Foo;  /* { dg-error "(parse|syntax) error before" } */
  
  double Bar;
  @interface Bar

Re: GCC-3.2.3 prerelease, 1st iteration

[Gabriel Dos Reis]

Ulrich Weigand <weigand@immd1.informatik.uni-erlangen.de> writes:

| Jakub Jelink wrote:
| 
| > On Sun, Apr 13, 2003 at 11:44:15PM +0200, Ulrich Weigand wrote:
| > > On both platforms the same set of (unexpected) regressions:
| > > FAIL: objc.dg/naming-1.m  (test for errors, line 20)
| > > FAIL: objc.dg/naming-1.m (test for excess errors)
| > > FAIL: objc.dg/naming-2.m  (test for errors, line 7)
| > > FAIL: objc.dg/naming-2.m (test for excess errors)
| > > 
| > > These appear to have crept into the 3.2 branch very recently,
| > > they were not present 20030329 when I last built it ...
| > 
| > Those are likely because you've upgraded bison.
| 
| Ah, I see.  However, it's not me who has upgraded bison,
| it must have been Gaby, because the prerelease tarball

I'm using the same environment as for 3.2.2.  I'm wondering why the
issue didn't show up for objc last time (I know it did for C and C++).

-- Gaby

Re: GCC-3.2.3 prerelease, 1st iteration

[Gabriel Dos Reis]

Jakub Jelinek <jakub@redhat.com> writes:

[...]

| > On both platforms the same set of (unexpected) regressions:
| > FAIL: objc.dg/naming-1.m  (test for errors, line 20)
| > FAIL: objc.dg/naming-1.m (test for excess errors)
| > FAIL: objc.dg/naming-2.m  (test for errors, line 7)
| > FAIL: objc.dg/naming-2.m (test for excess errors)
| > 
| > These appear to have crept into the 3.2 branch very recently,
| > they were not present 20030329 when I last built it ...
| 
| Those are likely because you've upgraded bison.
| 
| 2003-03-17  Zack Weinberg  <zack@codesourcery.com>
| 
|         * objc.dg/naming-1.m: Use "(parse|syntax) error".
|         * objc.dg/naming-2.m: Likewise.
| 
| should fix it, maybe it should be applied to gcc-3_2-branch too.

Yes, I backported it and am in the process of testing.  Then, I'll commit
and make a new prerelease.

Thanks,

-- Gaby

Re: GCC-3.2.3 prerelease, 1st iteration

[Christopher Faylor]

On Tue, Apr 15, 2003 at 06:07:38PM -0400, DJ Delorie wrote:
>>Thank you.  So, who is the official MinGW maintainer?  Christopher
>>Faylor?
>
>I think the way it works is, there are a few cygwin/mingw gcc folks,
>and either Chris or I just reviewed their patches and approved them.
>But I've been out of the cygwin loop long enough that I can't reliably
>review their patches any more.

I guess it's just me these days, then.

cgf

Re: GCC-3.2.3 prerelease, 1st iteration

[DJ Delorie]

> Thank you.  So, who is the official MinGW maintainer?  Christopher Faylor?

I think the way it works is, there are a few cygwin/mingw gcc folks,
and either Chris or I just reviewed their patches and approved them.
But I've been out of the cygwin loop long enough that I can't reliably
review their patches any more.

Re: GCC-3.2.3 prerelease, 1st iteration

[Gabriel Dos Reis]

DJ Delorie <dj@redhat.com> writes:

| >   windows, cygwin, mingw  DJ Delorie              dj@redhat.com
| > 
| > Is the first entry wrong?  If so, we need to update that file...
| 
| It's probably wrong now, but wasn't when it was added.  I'll change
| it.

Thank you.  So, who is the official MinGW maintainer?  Christopher Faylor?
 
-- Gaby

Re: GCC-3.2.3 prerelease, 1st iteration

[Stephan T. Lavavej]

Re. the problem compiling 3.2.3 on MinGW:

I've downloaded MinGW and MSYS to a fresh computer and I get the same
problem when compiling. So, it should be reproducible anywhere.

Stephan T. Lavavej
http://stl.caltech.edu

Re: GCC-3.2.3 prerelease, 1st iteration

[DJ Delorie]

>   windows, cygwin, mingw  DJ Delorie              dj@redhat.com
> 
> Is the first entry wrong?  If so, we need to update that file...

It's probably wrong now, but wasn't when it was added.  I'll change
it.

Re: GCC-3.2.3 prerelease, 1st iteration

[Gabriel Dos Reis]

DJ Delorie <dj@redhat.com> writes:

| > | I use MinGW, and I downloaded MSYS in order to compile the 3.2.3 prerelease.
| > | This is my first time compiling gcc.
| > | 
| > | jartool.c: In function `main':
| > | jartool.c:437: `S_IRGRP' undeclared (first use in this function)
| > | jartool.c:437: (Each undeclared identifier is reported only once
| > 
| > Hmm, something seems to be wrong with your system stat header file.
| > DJ, please, could you look into this?
| 
| Er, why me?  I normally don't have anything to do with the java
| sources or mingw...

I was naming you because our MAINTAINERS file says:

  dromion% grep DJ MAINTAINERS                          ~/redhat/3-2-x.gcc
  windows, cygwin, mingw  DJ Delorie              dj@redhat.com
  DJGPP                   DJ Delorie              dj@delorie.com
  libiberty               DJ Delorie              dj@redhat.com
  build machinery (*.in)  DJ Delorie              dj@redhat.com

Is the first entry wrong?  If so, we need to update that file...

Thanks,

-- Gaby
 

Re: GCC-3.2.3 prerelease, 1st iteration

[DJ Delorie]

> | I use MinGW, and I downloaded MSYS in order to compile the 3.2.3 prerelease.
> | This is my first time compiling gcc.
> | 
> | jartool.c: In function `main':
> | jartool.c:437: `S_IRGRP' undeclared (first use in this function)
> | jartool.c:437: (Each undeclared identifier is reported only once
> 
> Hmm, something seems to be wrong with your system stat header file.
> DJ, please, could you look into this?

Er, why me?  I normally don't have anything to do with the java
sources or mingw...

Re: GCC-3.2.3 prerelease, 1st iteration

[Gerald Pfeifer]

On Mon, 14 Apr 2003, Jakub Jelinek wrote:
> Those are likely because you've upgraded bison.
>
> 2003-03-17  Zack Weinberg  <zack@codesourcery.com>
>
>         * objc.dg/naming-1.m: Use "(parse|syntax) error".
>         * objc.dg/naming-2.m: Likewise.
>
> should fix it, maybe it should be applied to gcc-3_2-branch too.

Yes, that seems to be necessary.  Zack,would you mind doing that?

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

Re: GCC-3.2.3 prerelease, 1st iteration

[Ulrich Weigand]

Jakub Jelink wrote:

> On Sun, Apr 13, 2003 at 11:44:15PM +0200, Ulrich Weigand wrote:
> > On both platforms the same set of (unexpected) regressions:
> > FAIL: objc.dg/naming-1.m  (test for errors, line 20)
> > FAIL: objc.dg/naming-1.m (test for excess errors)
> > FAIL: objc.dg/naming-2.m  (test for errors, line 7)
> > FAIL: objc.dg/naming-2.m (test for excess errors)
> > 
> > These appear to have crept into the 3.2 branch very recently,
> > they were not present 20030329 when I last built it ...
> 
> Those are likely because you've upgraded bison.

Ah, I see.  However, it's not me who has upgraded bison,
it must have been Gaby, because the prerelease tarball
contains objc-parse.c generated with bison 1.875, while
I have bison 1.35 on my system ...

> 2003-03-17  Zack Weinberg  <zack@codesourcery.com>
> 
>         * objc.dg/naming-1.m: Use "(parse|syntax) error".
>         * objc.dg/naming-2.m: Likewise.
> 
> should fix it, maybe it should be applied to gcc-3_2-branch too.

Possibly.  On the other hand, this doesn't appear to be real problem,
so I don't really care ...

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  weigand@informatik.uni-erlangen.de

Re: GCC-3.2.3 prerelease, 1st iteration

[Jakub Jelinek]

On Sun, Apr 13, 2003 at 11:44:15PM +0200, Ulrich Weigand wrote:
> Gabriel Dos Reis wrote:
> 
> >  I uploaded GCC-3.2.3 prerelease tarballs on the FTP server as
> >
> >  ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*3.2.3-20030412.tar.*
> 
> Bootstrapped/regtested on s390-ibm-linux and s390x-ibm-linux,
> see reports at:
> http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg00800.html
> http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg00801.html
> 
> On both platforms the same set of (unexpected) regressions:
> FAIL: objc.dg/naming-1.m  (test for errors, line 20)
> FAIL: objc.dg/naming-1.m (test for excess errors)
> FAIL: objc.dg/naming-2.m  (test for errors, line 7)
> FAIL: objc.dg/naming-2.m (test for excess errors)
> 
> These appear to have crept into the 3.2 branch very recently,
> they were not present 20030329 when I last built it ...

Those are likely because you've upgraded bison.

2003-03-17  Zack Weinberg  <zack@codesourcery.com>

        * objc.dg/naming-1.m: Use "(parse|syntax) error".
        * objc.dg/naming-2.m: Likewise.

should fix it, maybe it should be applied to gcc-3_2-branch too.

	Jakub

Re: GCC-3.2.3 prerelease, 1st iteration

[Ulrich Weigand]

Gabriel Dos Reis wrote:

>  I uploaded GCC-3.2.3 prerelease tarballs on the FTP server as
>
>  ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*3.2.3-20030412.tar.*

Bootstrapped/regtested on s390-ibm-linux and s390x-ibm-linux,
see reports at:
http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg00800.html
http://gcc.gnu.org/ml/gcc-testresults/2003-04/msg00801.html

On both platforms the same set of (unexpected) regressions:
FAIL: objc.dg/naming-1.m  (test for errors, line 20)
FAIL: objc.dg/naming-1.m (test for excess errors)
FAIL: objc.dg/naming-2.m  (test for errors, line 7)
FAIL: objc.dg/naming-2.m (test for excess errors)

These appear to have crept into the 3.2 branch very recently,
they were not present 20030329 when I last built it ...

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  weigand@informatik.uni-erlangen.de

Re: GCC-3.2.3 prerelease, 1st iteration

[Gabriel Dos Reis]

"Stephan T. Lavavej" <stl@caltech.edu> writes:

| Hi,
| 
| I use MinGW, and I downloaded MSYS in order to compile the 3.2.3 prerelease.
| This is my first time compiling gcc.
| 
| ./configure ran fine, but make quickly failed, with this:
| 
| make[1]: Entering directory `/c/temp/gcc-3.2.3-20030412/fastjar'
| gcc -DHAVE_CONFIG_H -I. -I. -I. -I. -I. -I./../zlib -I./../include   -W
| -Wall -pedantic -Wstrict-prototypes -Wmissing-prototypes -Wwrite-strings -g
| -O2 -c jartool.c
| In file included from jartool.c:234:
| ../include/getopt.h:115: warning: function declaration isn't a prototype

I can't see any change to gcc-3.2.3/include/getopt.h between 3.2.1 and
3.2.2 and 3.2.3pre.  The most recent change to that file, according to
ChangeLog, is:

   2002-11-03  Roger Sayle  <roger@eyesopen.com>
               DJ Delorie <dj@redhat.com>

           * getopt.h:  Avoid prototyping getopt with no arguments in C++.

| jartool.c: In function `main':
| jartool.c:437: `S_IRGRP' undeclared (first use in this function)
| jartool.c:437: (Each undeclared identifier is reported only once

Hmm, something seems to be wrong with your system stat header file.
DJ, please, could you look into this?


[...]

| I don't know if this is a problem with gcc or on my end.  It seems to me
| that ./configure followed by make should "just work", though.

That is the intent.  Thanks for your report.

-- Gaby

Re: GCC-3.2.3 prerelease, 1st iteration

[Stephan T. Lavavej]

Hi,

I use MinGW, and I downloaded MSYS in order to compile the 3.2.3 prerelease.
This is my first time compiling gcc.

./configure ran fine, but make quickly failed, with this:

make[1]: Entering directory `/c/temp/gcc-3.2.3-20030412/fastjar'
gcc -DHAVE_CONFIG_H -I. -I. -I. -I. -I. -I./../zlib -I./../include   -W
-Wall -pedantic -Wstrict-prototypes -Wmissing-prototypes -Wwrite-strings -g
-O2 -c jartool.c
In file included from jartool.c:234:
../include/getopt.h:115: warning: function declaration isn't a prototype
jartool.c: In function `main':
jartool.c:437: `S_IRGRP' undeclared (first use in this function)
jartool.c:437: (Each undeclared identifier is reported only once
jartool.c:437: for each function it appears in.)
jartool.c:437: `S_IROTH' undeclared (first use in this function)
jartool.c: In function `extract_jar':
jartool.c:1427: too many arguments to function `mkdir'
jartool.c: In function `list_jar':
jartool.c:1682: warning: pointer targets in passing arg 1 of
`__builtin_puts' differ in signedness
jartool.c:1689: warning: pointer targets in passing arg 1 of
`__builtin_puts' differ in signedness
jartool.c:1827: warning: pointer targets in passing arg 1 of
`__builtin_puts' differ in signedness
jartool.c:1834: warning: pointer targets in passing arg 1 of
`__builtin_puts' differ in signedness
jartool.c: In function `help':
jartool.c:1893: warning: string length `594' is greater than the length
`509' ISO C89 compilers are required to support
jartool.c:1893: warning: string length `593' is greater than the length
`509' ISO C89 compilers are required to support
make[1]: *** [jartool.o] Error 1
make[1]: Leaving directory `/c/temp/gcc-3.2.3-20030412/fastjar'
make: *** [all-fastjar] Error 2

I don't know if this is a problem with gcc or on my end.  It seems to me
that ./configure followed by make should "just work", though.

Stephan T. Lavavej
http://stl.caltech.edu