GCC 3.3 Prerelease

GCC 3.3 Prerelease

[Mark Mitchell]

The first GCC 3.3 prerelease is available on gcc.gnu.org:

  ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*-3.3-20030506.tar.gz

The final GCC 3.3 release will contain very few changes, relative to
this code base, but it is likely that we will fix a few important bugs,
configuration problems, and/or packaging problems.

If you find problems in this pre-release, do not send them directly to
me.  Instead, please file a bug-report here:

  http://gcc.gnu.org/cgi-bin/gnatsweb.pl

and add my name to the CC list for the bug report.

If you do send me a report directly, I will simply return it to you with
a request to put it in GNATS, as above.

Yours,

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com

Re: GCC 3.3 Prerelease

[Joe Buck]

On Tue, May 06, 2003 at 09:17:27AM -0700, Mark Mitchell wrote:
> The first GCC 3.3 prerelease is available on gcc.gnu.org:
> 
>   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*-3.3-20030506.tar.gz

The file gcc-3.3-2030506.tar.gz is only 128 bytes long, so is
gcc-3.3-core-3.3-2030506.tar.gz.  I think you'll need to try again.
 

Re: GCC 3.3 Prerelease

[Mark Mitchell]

On Tue, 2003-05-06 at 09:33, Joe Buck wrote:
> On Tue, May 06, 2003 at 09:17:27AM -0700, Mark Mitchell wrote:
> > The first GCC 3.3 prerelease is available on gcc.gnu.org:
> > 
> >   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*-3.3-20030506.tar.gz

Remember how I said "and correct packaging problems?"

Well, we've got one -- some of the prerelease tarfiles are empty.

New tarfiles will be available as fast as the network can copy them
there.

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com

Re: GCC 3.3 Prerelease

[Gerald Pfeifer]

[ gcc-announce removed ]

On Tue, 6 May 2003, Mark Mitchell wrote:
> New tarfiles will be available as fast as the network can copy them
> there.

Thanks!  For releases and regular snapshots we also used to have .bz2
files; did you omit them intentionally for this prerelease or is this
a problem with the script?

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

Re: GCC 3.3 Prerelease

[Paul Jarc]

Mark Mitchell <mark@codesourcery.com> wrote:
>   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*-3.3-20030506.tar.gz

Some people have asked that the version string be 3.3.0 instead of
3.3.  I never saw any objections to this, and I'd like to see it too.
It helps to remove the ambiguity in "3.3" (was the speaker referring
to all 3.3.* versions, or just the first?), and also helps sorting:
$ ls gcc-3.3*
gcc-3.3.0.tar.gz
gcc-3.3.1.tar.gz

versus:
$ ls gcc-3.3*
gcc-3.3.1.tar.gz
gcc-3.3.tar.gz


paul

Re: GCC 3.3 Prerelease

[Mark Mitchell]

On Tue, 2003-05-06 at 10:00, Gerald Pfeifer wrote:
> [ gcc-announce removed ]
> 
> On Tue, 6 May 2003, Mark Mitchell wrote:
> > New tarfiles will be available as fast as the network can copy them
> > there.
> 
> Thanks!  For releases and regular snapshots we also used to have .bz2
> files; did you omit them intentionally for this prerelease or is this
> a problem with the script?

I omitted them intentionally; it takes long enough to upload the files
as it is...

I will make sure that they, along with the diffs, are there in the final
release.

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com

Re: GCC 3.3 Prerelease

[Mark Mitchell]

On Tue, 2003-05-06 at 10:32, Paul Jarc wrote:
> Mark Mitchell <mark@codesourcery.com> wrote:
> >   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*-3.3-20030506.tar.gz
> 
> Some people have asked that the version string be 3.3.0 instead of
> 3.3.  I never saw any objections to this, and I'd like to see it too.

I don't object either, but AFAIK, ".0" is not the usual GNU way.

We'd need a patch for GCC, and approval from the FSF.  We might do this
for 3.4[.0] -- but 3.3 will be 3.3.

Thanks,

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com

Re: GCC 3.3 Prerelease

[Joe Buck]

On Tue, May 06, 2003 at 11:53:18AM -0700, Mark Mitchell wrote:
> On Tue, 2003-05-06 at 10:32, Paul Jarc wrote:
> > Mark Mitchell <mark@codesourcery.com> wrote:
> > >   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*-3.3-20030506.tar.gz
> > 
> > Some people have asked that the version string be 3.3.0 instead of
> > 3.3.  I never saw any objections to this, and I'd like to see it too.
> 
> I don't object either, but AFAIK, ".0" is not the usual GNU way.
> 
> We'd need a patch for GCC, and approval from the FSF.  We might do this
> for 3.4[.0] -- but 3.3 will be 3.3.

I don't think that we need approval from the FSF on something as trivial
as that; the SC could decide it.

I agree that it is too late to change 3.3 after a prerelease tarball has
already gone out. 

Re: GCC 3.3 Prerelease

[Joe Buck]

On Tue, May 06, 2003 at 09:17:27AM -0700, Mark Mitchell wrote:
> The first GCC 3.3 prerelease is available on gcc.gnu.org:
> 
>   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*-3.3-20030506.tar.gz

Test results for a Red Hat 7.2 system with updated binutils are at

http://gcc.gnu.org/ml/gcc-testresults/2003-05/msg00361.html
 

Re: GCC 3.3 Prerelease

[David Edelsohn]

>>>>> Joe Buck writes:

Joe> I don't think that we need approval from the FSF on something as trivial
Joe> as that; the SC could decide it.

	I don't think that GCC should shift away from GNU Project
precedent without asking.

David

Re: GCC 3.3 Prerelease

[Mark Mitchell]

On Tue, 2003-05-06 at 12:23, Joe Buck wrote:
> On Tue, May 06, 2003 at 09:17:27AM -0700, Mark Mitchell wrote:
> > The first GCC 3.3 prerelease is available on gcc.gnu.org:
> > 
> >   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*-3.3-20030506.tar.gz
> 
> Test results for a Red Hat 7.2 system with updated binutils are at
> 
> http://gcc.gnu.org/ml/gcc-testresults/2003-05/msg00361.html

Thanks.  The unused-4.c failure is a CVS mis-operation on my part; after
fixing the rcs_id problem I failed to commit the testcase.  That's now
fixed.  Other results are as expected.

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com

Re: GCC 3.3 Prerelease

[Joe Buck]

On Tue, May 06, 2003 at 12:39:20PM -0700, Mark Mitchell wrote:
> On Tue, 2003-05-06 at 12:23, Joe Buck wrote:
> > On Tue, May 06, 2003 at 09:17:27AM -0700, Mark Mitchell wrote:
> > > The first GCC 3.3 prerelease is available on gcc.gnu.org:
> > > 
> > >   ftp://gcc.gnu.org/pub/gcc/snapshots/gcc-*-3.3-20030506.tar.gz
> > 
> > Test results for a Red Hat 7.2 system with updated binutils are at
> > 
> > http://gcc.gnu.org/ml/gcc-testresults/2003-05/msg00361.html
> 
> Thanks.  The unused-4.c failure is a CVS mis-operation on my part; after
> fixing the rcs_id problem I failed to commit the testcase.  That's now
> fixed.  Other results are as expected.

I now have a result for hppa2.0w-hp-hpux11.00, see

http://gcc.gnu.org/ml/gcc-testresults/2003-05/msg00366.html

The first g++ failure is an ICE.  From the g++.log file:

bitfield7_x.C: In function `void bitfield7_x()':
/remote/dtg103/jbuck/gnu/src/gcc-3.3-20030506/gcc/testsuite/g++.dg/compat/break/bitfield7_x.C:13: error: unrecognizable insn:
(insn 14 13 15 0 680a7c30 (set (zero_extract:SI (reg:SI 94)
            (const_int 32 [0x20])
            (const_int 0 [0x0]))
        (const_int 7 [0x7])) -1 (nil)
    (nil))
/remote/dtg103/jbuck/gnu/src/gcc-3.3-20030506/gcc/testsuite/g++.dg/compat/break/bitfield7_x.C:13: internal compiler error: in extract_insn, at recog.c:2175
Please submit a full bug report,

Is this a known issue?  Other than that, things look fine.
 

Re: GCC 3.3 Prerelease

[Gerald Pfeifer]

On Tue, 6 May 2003, Mark Mitchell wrote:
>> Thanks!  For releases and regular snapshots we also used to have .bz2
>> files; did you omit them intentionally for this prerelease or is this
>> a problem with the script?
> I omitted them intentionally; it takes long enough to upload the files
> as it is...

I just noticed because it broke a tester of mine and the FreeBSD Ports
Collection also needed changes, as both relied on the .bz2s.

If it's really inconvenient for you to upload both .gz and .bz2, effective
immediately, generating .bz2 (or checking whether they exist) is as easy as

  ssh mmitchel@gcc.gnu.org '~gerald/gz2bz2'

:-)

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

Re: GCC 3.3 Prerelease

[Mark Mitchell]

On Thu, 2003-05-08 at 02:29, Gerald Pfeifer wrote:
> On Tue, 6 May 2003, Mark Mitchell wrote:
> >> Thanks!  For releases and regular snapshots we also used to have .bz2
> >> files; did you omit them intentionally for this prerelease or is this
> >> a problem with the script?
> > I omitted them intentionally; it takes long enough to upload the files
> > as it is...
> 
> I just noticed because it broke a tester of mine and the FreeBSD Ports
> Collection also needed changes, as both relied on the .bz2s.

Sorry about that; I'll generate them in future.

Thanks!

-- 
Mark Mitchell <mark@codesourcery.com>
CodeSourcery, LLC

Re: GCC 3.3 Prerelease

[Laurent GUERBY]

FYI, gcc-3.3-20030506.tar.gz with c,ada works fine ACATS-wise on Red Hat
Linux 9 x86 using the system compiler (3.2.2-5) to bootstrap.

Laurent

Re: GCC 3.3 Prerelease

[<yunakaof-AT-usp.br>]

FYI,

the snapshot
 gcc-3.3-20030506.tar.gz

could be compiled well, and could be used with gcc and g++
to compile other gnu packages like courier-imap 1.7.1 that
use both languages.

I have

# more /etc/motd
Compaq Tru64 UNIX V5.1B (Rev. 2650); Thu Apr  3 17:53:31 BRT 2003
Tru64 UNIX Catalan Support V5.1B (rev. 232)
Tru64 UNIX Czech Support V5.1B (rev. 232)
Tru64 UNIX Hungarian Support V5.1B (rev. 232)
Tru64 UNIX Lithuanian Support V5.1B (rev. 232)
Tru64 UNIX Polish Support V5.1B (rev. 232)
Tru64 UNIX Russian Support V5.1B (rev. 232)
Tru64 UNIX Slovak Support V5.1B (rev. 232)
Tru64 UNIX Slovene Support V5.1B (rev. 232)
Tru64 UNIX Thai Support V5.1B (rev. 232)
Tru64 UNIX French Support V5.1B (rev. 232)
Tru64 UNIX German Support V5.1B (rev. 232)
Tru64 UNIX Greek Support V5.1B (rev. 232)
Tru64 UNIX Hebrew Support V5.1B (rev. 232)
Tru64 UNIX Italian Support V5.1B (rev. 232)
Tru64 UNIX Japanese Support V5.1B (rev. 232)
Tru64 UNIX Korean Support V5.1B (rev. 232)
Tru64 UNIX Spanish Support V5.1B (rev. 232)
Tru64 UNIX Swedish Support V5.1B (rev. 232)
Tru64 UNIX Chinese Support V5.1B (rev. 232)
Tru64 UNIX Turkish Support V5.1B (rev. 232)

with alpha processor ( ex-DEC, ex-Compaq, now HP DS20 workstation )

# /usr/sbin/psrinfo -v
Status of processor 0 as of: 05/08/03 21:51:54
  Processor has been on-line since 04/25/2003 16:11:14
  The alpha EV6.7 (21264A) processor operates at 667 MHz,
  has a cache size of 8388608 bytes,
  and has an alpha internal floating point processor.

Status of processor 1 as of: 05/08/03 21:51:54
  Processor has been on-line since 04/25/2003 16:11:14
  The alpha EV6.7 (21264A) processor operates at 667 MHz,
  has a cache size of 8388608 bytes,
  and has an alpha internal floating point processor.

another languages, that are build with

make bootstrap4

were apparently successfull.

I have previously installed some GNU packages like
texinfo
m4
autoconf
automake
bison
bash
gawk
gettext
libconv
make
and only "strip" from binutils
where I make via PATH that for GNU
compilation is used this. ( I am not "make install" this )

I used the native /usr/ccs/bin/cc
and with "Development Tool Kit License" license
installed in system.

I could not make the check, cause I could not successfully
compile expect in this OS.


Nelson



On 8 May 2003, Laurent GUERBY wrote:

%%  Date: 08 May 2003 22:16:30 +0200
%%  From: Laurent GUERBY <guerby@acm.org>
%%  To: Mark Mitchell <mark@codesourcery.com>
%%  Cc: gcc@gcc.gnu.org
%%  Subject: Re: GCC 3.3 Prerelease
%%
%%  FYI, gcc-3.3-20030506.tar.gz with c,ada works fine ACATS-wise on Red Hat
%%  Linux 9 x86 using the system compiler (3.2.2-5) to bootstrap.
%%
%%  Laurent
%%
%%

Re: GCC 3.3 Prerelease

[Phil Edwards]

On Thu, May 08, 2003 at 11:29:27AM +0200, Gerald Pfeifer wrote:
> On Tue, 6 May 2003, Mark Mitchell wrote:
> >> Thanks!  For releases and regular snapshots we also used to have .bz2
> >> files; did you omit them intentionally for this prerelease or is this
> >> a problem with the script?
> > I omitted them intentionally; it takes long enough to upload the files
> > as it is...
> 
> I just noticed because it broke a tester of mine and the FreeBSD Ports
> Collection also needed changes, as both relied on the .bz2s.
> 
> If it's really inconvenient for you to upload both .gz and .bz2,

then perhaps he shouldn't have to.

Given:  - bz2 files are smaller, therefore faster to upload,
        - more people are depending on the bz2s,
        - duplicating formats is trivial

Why not have a script on ftp.gnu.org that a RM can trigger after uploading
the bz2 files which creates gz equivalents?


Phil

-- 
If ye love wealth greater than liberty, the tranquility of servitude greater
than the animating contest for freedom, go home and leave us in peace.  We seek
not your counsel, nor your arms.  Crouch down and lick the hand that feeds you;
and may posterity forget that ye were our countrymen.            - Samuel Adams

Re: GCC 3.3 Prerelease

[Gerald Pfeifer]

On Thu, 8 May 2003, Mark Mitchell wrote:
>> I just noticed because it broke a tester of mine and the FreeBSD Ports
>> Collection also needed changes, as both relied on the .bz2s.
> Sorry about that; I'll generate them in future.

Thanks!

On Thu, 8 May 2003, Phil Edwards wrote:
>> If it's really inconvenient for you to upload both .gz and .bz2,
> then perhaps he shouldn't have to.
>
> Given:  - bz2 files are smaller, therefore faster to upload,
>         - more people are depending on the bz2s,
>         - duplicating formats is trivial
>
> Why not have a script on ftp.gnu.org that a RM can trigger after
> uploading the bz2 files which creates gz equivalents?

This is exactly what I had written for Mark, alas on gcc.gnu.org!

Mark, did you use my script to generate the .bz2 files for the second
prerelease? Respectively, would you like me to adjust it to convert from
.bz2 to .gz?

If that helps reduce your load (or the one of your uplink), I'll happily
do whatever you prefer. :-)

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

Re: GCC 3.3 Prerelease

[Mark Mitchell]

> If that helps reduce your load (or the one of your uplink), I'll happily
> do whatever you prefer. :-)

I really appreciate your help.  It's really not a big deal; I just tried
to cut corners the first time, and now that I know people care I won't
do that in the future.

Thanks a lot for being so helpful!

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com

Re: GCC 3.3 Prerelease

[Mark Mitchell]

> bitfield7_x.C: In function `void bitfield7_x()':
> /remote/dtg103/jbuck/gnu/src/gcc-3.3-20030506/gcc/testsuite/g++.dg/compat/break/bitfield7_x.C:13: error: unrecognizable insn:
> (insn 14 13 15 0 680a7c30 (set (zero_extract:SI (reg:SI 94)
>             (const_int 32 [0x20])
>             (const_int 0 [0x0]))
>         (const_int 7 [0x7])) -1 (nil)
>     (nil))
> /remote/dtg103/jbuck/gnu/src/gcc-3.3-20030506/gcc/testsuite/g++.dg/compat/break/bitfield7_x.C:13: internal compiler error: in extract_insn, at recog.c:2175
> Please submit a full bug report,
> 
> Is this a known issue?  Other than that, things look fine.

Yes, this is a known issue.  Some platforms have problems with that
test, due to it's use of a 4096-bit bitfield.  This is an extreme corner
case.  There is a GNATS entry for the PowerPC version of this failure:
PR 8306.

I've updated that to mention that HP-PA fails too.

Thanks,

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com

Re: GCC 3.3 Prerelease

[Jeff Bailey]

On Tue, 2003-05-06 at 11:53:18 (-0700), Mark Mitchell wrote:

> We'd need a patch for GCC, and approval from the FSF.  We might do
> this for 3.4[.0] -- but 3.3 will be 3.3.

Taking a quick look at ftp.gnu.org, I see that inetutils and lilypond
have x.y.0 version numbers, but that wget and gnussl do not.

There was some discussion back on the GNU maintainers list about how
version numbers should be handled, but there was no concensus.  The GCS
hasn't been updated to reflect any sort of decision.

Tks,
Jeff Bailey

-- 
You said homosexuals form a small percentage of the population.  So
do Jews.  Is that a reason to deny someone equality?
 - Richard Marceau