ping/summary of recent testsuite patches

[Janis Johnson <janis187 at us dot ibm dot com>]

I'd like to summarize the testsuite patches I've submitted recently and
explain how they fit together.

http://gcc.gnu.org/ml/gcc-patches/2004-11/msg00154.html

  Fixes dg-xfail-if to prevent it from setting any xfail flags for a
  test that will be skipped because of a dg-do target list or
  dg-require*.  Without this fix the test that is run next gets that
  flag and is reported as XPASS rather than PASS

http://gcc.gnu.org/ml/gcc-patches/2004-11/msg00154.html

  Fixes existing dg-require-* procs to clear out xfail flags that were
  set for an earlier dg-do xfail list or dg-xfail-if in the same test.
  Without this fix the test that is run next gets that flag and is
  reported as XPASS rather than PASS.

http://gcc.gnu.org/ml/gcc-patches/2004-11/msg00172.html

  Adds new procs
    check_effective_target_ilp32
    check_effective_target_lp64
    check_effective_target
    dg-require-effective-target
  and uses them in some powerpc64 tests that should not be run for -m32.
  The latter two procs take a keyword argument, but the bigger purpose
  I had in mind for the keyword is no longer needed, so they could be
  replaced with dg-require-effective-target-ilp32 and
  dg-require-effective-target-lp64 if that seems cleaner.

http://gcc.gnu.org/ml/gcc-patches/2004-11/msg00417.html

  Adds new proc dg-target-list, which adds flexibility to target and
  xfail lists within dg- commands.

http://gcc.gnu.org/ml/gcc-patches/2004-11/msg00156.html

  Adds support for testing the test framework.  This version depends
  only on a new proc from the dg-require-* patch and demonstrates the
  problems fixed in that patch and the dg-xfail-if patch.  The patch
  would put some files into contrib/test_framework, although I've also
  thought about having the test generator in the testsuite, ignored
  unless an environment variable is set.

  I've got a new version of these tests that exercises dg-target-list
  that I can support if there's interest.  I don't really care if this
  goes in, mostly it's just to demonstrate how I've been testing these
  changes.

There are also some unreviewed patches that change particular tests or
groups of tests:

http://gcc.gnu.org/ml/gcc-patches/2004-10/msg02630.html

  Cleans up gcc.misc-tests/linkage.c by adding the directory name to
  pass/fail messages and removing temporary files.

http://gcc.gnu.org/ml/gcc-patches/2004-10/msg02631.html

  Allows gcc.misc-tests/linkage.c to pass when a powerpc64*-*-linux*
  compiler is tested with -m64.

http://gcc.gnu.org/ml/gcc-patches/2004-10/msg01746.html

  Cleans up the auto-vectorization tests in gcc.dg/vect, simplifying
  the test headers and allowing the tests to be compile-only on a
  system that supports compiling, but not executing, vector code.  This
  is based on a patch that Aldy submitted.  With the new framework
  support I'd like to re-do this patch to be even cleaner.

Janis