GCC 3.0 Status

GCC 3.0 Status

[Mark Mitchell]

I have created a (very ugly) web page describing the status of the GCC
3.0 branch.  (There is no branch yet.)

This page gives the target triplets and volunteers for the
stabilization testing we are doing this week, as well as directions
for how to perform that stabilization testing.  Once people have run
some tests, the test status will be there too.

Right now, some of our target platforms are still unrepresented.  For
example, there is no AIX volunteer, and no HPUX volunteer.  If you
have access to an a machine for one of these targets, and would be
willing to test, please send me mail.

The web page is at:

  http://gcc.gnu.org/gcc-3.0/branch.html

Thank you!

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

Re: GCC 3.0 Status

[Joe Buck]

> Right now, some of our target platforms are still unrepresented.  For
> example, there is no AIX volunteer, and no HPUX volunteer.  If you
> have access to an a machine for one of these targets, and would be
> willing to test, please send me mail.

I have an old HPUX 10.20 machine, which I just tried.  Currently things
are badly broken and the compiler won't build, see recent gcc-bugs
report.

Re: GCC 3.0 Status

[Jeffrey A Law]

  In message < 200101161831.KAA17866@racerx.synopsys.com >you write:
  > 
  > > Right now, some of our target platforms are still unrepresented.  For
  > > example, there is no AIX volunteer, and no HPUX volunteer.  If you
  > > have access to an a machine for one of these targets, and would be
  > > willing to test, please send me mail.
  > 
  > I have an old HPUX 10.20 machine, which I just tried.  Currently things
  > are badly broken and the compiler won't build, see recent gcc-bugs
  > report.
I'm working on the codegen problem.  It's related to some recent changes.
I also know what the gcc2_compiled problem is.  That'll take about 30
seconds to fix :-)

jeff

Re: GCC 3.0 Status

[Jim Wilson]

IA-64 isn't one of the primary or secondary platforms, but it would be nice
if it would build.  IA-64 builds have been failing ever since we switched to
libstdc++-v3 because a C++ front end change broke machine dependent builtin
functions.  I reported this in November.
	http://gcc.gnu.org/ml/gcc-bugs/2000-11/msg00849.html
There was a proposed patch in December from Herman ten Brugge
	http://gcc.gnu.org/ml/gcc-patches/2000-12/msg00596.html
which hasn't been reviewed yet.

I have time to work on the current gcc sources this week, so I am hoping to
get this fixed before we branch.  I'll also clean up some other issues, such
as documenting the IA-64 options.

Previously I was busy stabilizing a late summer snapshot enough so that OS
releases could be made, and that work finished up in December before I left
on vacation.  So this is my first chance to work on this problem.

Jim

Re: GCC 3.0 Status

[Richard Henderson]

On Tue, Jan 16, 2001 at 12:24:00PM -0800, Jim Wilson wrote:
> IA-64 isn't one of the primary or secondary platforms, but it would be nice
> if it would build.  IA-64 builds have been failing ever since we switched to
> libstdc++-v3 because a C++ front end change broke machine dependent builtin
> functions.  I reported this in November.

Theoretically this was fixed here:

2001-01-10  Mark Mitchell  <mark@codesourcery.com>

        * c-common.h (CTI_VOID_LIST): Remove.
        (void_list_node): Likewise.
        * tree.h (TI_VOID_LIST_NODE): New enumeral.
        (void_list_node): New macro.
        * config/arm/arm.c (arm_init_builtins): Use void_list_node.
        * config/i386/i386.c (ix86_init_builtins): Likewise.
        * config/ia64a/ia64.c (ia64_init_builtins): Likewise.

I havn't run an ia64 bootstrap since then to verify though.


r~

Re: GCC 3.0 Status

[Mark Mitchell]

>>>>> "Joe" == Joe Buck <jbuck@racerx.synopsys.com> writes:

    >> Right now, some of our target platforms are still
    >> unrepresented.  For example, there is no AIX volunteer, and no
    >> HPUX volunteer.  If you have access to an a machine for one of
    >> these targets, and would be willing to test, please send me
    >> mail.

    Joe> I have an old HPUX 10.20 machine, which I just tried.
    Joe> Currently things are badly broken and the compiler won't
    Joe> build, see recent gcc-bugs report.

Thanks for both bits of information.

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

Re: GCC 3.0 Status

[Mark Mitchell]

>>>>> "Jeffrey" == Jeffrey A Law <law@redhat.com> writes:

    Jeffrey> I'm working on the codegen problem.  It's related to some
    Jeffrey> recent changes.  I also know what the gcc2_compiled
    Jeffrey> problem is.  That'll take about 30 seconds to fix :-)

Thanks, and yay!

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

GCC 3.0 Status

[Mark Mitchell]

Here's my weekly GCC 3.0 Status Report.

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

GCC 3.0 Status Report
=====================

Last Week
---------

  The major progress for the week was in the area of analyzing the
  open bugs.  We now have volunteers for 509 of the 958 bugs that 
  were originally open.  897 bugs remain open, meaning that a
  significant number were either already fixed, or have been
  fixed during this process.

  Progress was slowed by the fact that the 3.0 branch would not build
  on x86 for a couple of days.  Please take care to bootstrap and
  test all changes on the branch.  This is of course required on the
  mainline as well, but it is even more important on the branch.

  There were some compile-time performance improvements originating
  from Jakub Jelinek and Jeffrey Oldham.  

  Jeffrey Oldham also began preparing the POOMA testkit for GCC 3.0.
  He has verified that G++ compiles the sample programs that will be
  used, and is preparing a script to automatically build them, run
  them, and verify the results.

  Of course, other bugs were stomped.

Kudos
-----

  The GCC 3.0 Volunteer of the Week (an award that I just made up,
  and which comes with absolutely no tangible reward of any kind) is a
  contributor whose work on the preprocessor is of course legendary.
  This week, however, he showed his willingness to plunge into 
  the relatively thankless job of bug reviewing, volunteering to
  review an astounding 113 bugs!

  Ladies and gentlemen, our volunteer of the week is...

  ...Neil Booth!

  Congratulations!

Next Week
---------

  Continue analyzing open bugs.

  Begin working on compile-time performance issues in earnest.

GCC 3.0 Status

[Mark Mitchell]

At our company meeting this morning, it was suggested that I send out
a weekly announcement regarding the state of GCC 3.0.  I think that's
a terrific idea, so here's the first status report.  I'll try to keep
these coming until the release is out the door.

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

GCC 3.0 Status Report
=====================

Overall
-------

  We have branched for the release.  We believe that most (if not all)
  necessary functionality is provided on the 3.0 release branch.
  
  At this time, the focus is on analyzing reports filed in the GNATS 
  database to find regressions from GCC 2.95.2.  Any such regressions
  must be fixed before the release.

Bug Status
----------

  As of a week ago, there were 958 open bugs.  There are now 928 open
  bugs.

  We have volunteers to look at 228 of these 958 bugs.  Clearly, we
  need more volunteer assistance.  Many of the remaining bugs are C++
  bugs, but many of them are not.  If you have not yet volunteered,
  please read:
   
    http://gcc.gnu.org/ml/gcc/2001-02/msg00405.html

  Then, to volunteer, please send mail to `mark@codesourcery.com'.  I
  am maintaining a master list of the bugs that have been processed
  so that we can avoid duplication.

Tasks Remaining
---------------

Create Release Branch:
  
  Completed 2000-02-11

Review GNATS Reports:

  Underway.

Improve Compile-time Performance:

  Underway.

Perform Application Testing:

  Not started.

Perform Benchmark Testing:

  Not started.

Prepare Release Notes:
 
  Not started.

Prepare Release Candidates:

  Not started.

Re: GCC 3.0 Status

[Janis Johnson]

> IA-64 isn't one of the primary or secondary platforms, but it would be
nice
> if it would build.  ...
>
> I have time to work on the current gcc sources this week, so I am hoping
to
> get this fixed before we branch.  I'll also clean up some other issues,
such
> as documenting the IA-64 options.

I'd like to build the latest gcc sources on an IA-64 platform and help
track down problems there.  I've got access to a development system that
is currently running Red Hat Linux 7.0 plus patches, but could install
some other version of GNU/Linux if that would make it easier to build
gcc.

Jim, how far should a bootstrap get with the current sources, and what
configuration options are needed to build at least the C compiler?  Are
there patches available to build dejagnu on IA-64?

Janis