Current failures in the gfortran testsuite...

[François-Xavier Coudert]

Hi all,

Apart from the do_3.F90 issue, there are currently a few failures
showing up in our testsuite on non-mainstream targets. (In particular,
Matthias Klose has recently posted a large number of linux testresults
on lots of processors.) The following points are to be noted:

  * gfortran.dg/deallocate_stat.f90 ICEs at -O1 on sparc-linux. This
worries, as this might mean we are emitting wrong code on other
platforms. Matthias, could you please file a PR for this? I don't
think anyone in the gfortran team has a sparc-linux access somewhere,
but hopefully we can still try to figure it out.

  * gfortran.dg/char_pointer_func.f90 is failing at -O3 on mips-linux
and mipsel-linux. Matthias, if you can open a PR with info, thanks
again!

  * alpha-linux and powerpc-aix are a bit of a mess. In particular,
they reveal bugs in the integer exponentiation routines
(http://gcc.gnu.org/ml/gcc-testresults/2007-09/msg00881.html and
http://gcc.gnu.org/ml/gcc-testresults/2007-09/msg00863.html)

  * hppa1.1-hp-hpux10.20 and hppa64-hp-hpux11.11 have also a fair
number of failures; I'll try to work on these (thanks to
testdrive.hp.com)


FX


PS: if you feel that my "regular" gcc-testresults watchdog mails are
not necessary, or counter-productive, please let me know.