Re: GCC 3.4 Release Status (2004-01-05)

[Andrew Pinski <pinskia at physics dot uc dot edu>]

On Jan 7, 2004, at 20:18, Giovanni Bajo wrote:
> I would like to put an emphasis on the testsuite results. Results are 
> not very
> appealing on some of the primary targets:

Eric already answered this one:
> mips-sgi-irix: 
> http://gcc.gnu.org/ml/gcc-testresults/2004-01/msg00302.html

I will talk about the rest.

For this one:
> powerpc-ibm-aix: 
> http://gcc.gnu.org/ml/gcc-testresults/2004-01/msg00309.html
Note I am comparing this to 
<http://gcc.gnu.org/ml/gcc-testresults/2003-01/msg01420.html>.
intrinsic77.f was not failing last year so it is a regression.
g++.old-deja/g++.abi/ptrflags.C was not failing last year.
g++.old-deja/g++.eh/catchptr1.C: Likewise

For this one:
> sparc-sun-solaris: 
> http://gcc.gnu.org/ml/gcc-testresults/2004-01/msg00289.html
Really there is no regressions at all except maybe in libjava.
All the gcc.dg/compat/ tests are new and so are the failing objective-C 
ones
(note these are due to 64 bit vs 32 bit and ABI differences between 
where the
tests were written for, the tests are wrong).
The failures in libffi are not regressions at all and they are also new 
tests anyway
and this is being worked on.
This looks bad because of libffi and there is a multi-libed target so 
there are twice
as many problems that could happen.