egcs-1.1 release issues

[Jeffrey A Law <law at cygnus dot com>]

Here's a quick update for what still needs to be done:

Regression testing:

  There's still a few regressions on the rs6000/powerpc AIX targets.
  For various reasons I can't perform any tests on them right now,
  so it is important for someone to test them with the latest
  snapshot.

  There's still some questions about some irix5.2 failures, but they
  are starting to look like hardware failures.

  We never got sparc-linux test results.  If they do not appear soon
  we'll have to drop it off the test matrix.

  The rest of the targets look good.

libg++ testing

  When you test the next snapshot, please be sure to include libg++
  tests.  Basically all you should need to do is get the libg++
  add-on, drop it into your egcs source tree and build/test normally.
  You can get the libg++ add-on on the egcs ftp server in the
  pub/egcs/infrastructure directory.

  I do not believe the standard reporting script looks for libg++
  results, so please explicitly run "make check" in the libg++ objdir
  and attach the results.


glibc testing

  I've placed a tarball of the glibc sources we're using for testing
  purposes.  You're supposed to configure (see instructions in glibc)
  then "make && make check".  Note you will need to ftp the crypt
  add-on from an appropriate location (should be mentioned in the
  INSTALL file).

  We need these test results for the x86, alpha, sparc and powerpc.

  Manfred reports the his glibc testing on the x86 has been successful,
  but he might have been testing a different version than we wanted to
  test.  Similarly for Andreas Jaeger on the x86.  I'm personally testing
  from the tarball right now on the x86.

  Similarly for Joel Klecker and Franz Sirl have been testing glibc on
  powerpcs.  However, Geoff Keating found a problem and has sent in
  a bug+fix that has not been evaluated yet.


linux kernel 2.1.xxx

   I'm unsure how to proceed with this one since apparently certain
   versions of the kernel have been rather unstable.  Can those folks
   who regularly follow kernel development recommend particular kernel
   versions for testing on the x86, alpha, sparc and powerpc?

   Manfred reports that he's been having success with 2.0.xx and 2.1.xxx
   kernels on the x86.

   Franz Sirl reports success with 2.1.115 on the powerpc.


redhat builds

  The RedHat 5.1 builds for the x86 have gone quite well.  There are
  a few packages that had invalid C++ code and such, but otherwise
  it was a reasonably smooth process on the x86.


Executive summary.

  There's a couple minor regression testing issues pending resolution.

  We need libg++ testing across the board.

  We need sparc & alpha testing for glibc, the linux kernel and the
  RedHat 5.1 SRPMS.
  
  There will have to be at least one more snapshot to pick up some
  rs6000/powerpc fixes, documentation updates and any fixes we need
  based on problems found with the remaining testing issues.