Re: (toplevel) Fix dramatic breakage for ordinary crosses (related to program_transform_name)

[Ben Elliston <bje@redhat.com>]

>>>>> "Michael" == Michael Elizabeth Chastain <mec@shout.net> writes:

  Michael> I would also prefer no locks, because I don't want to deal
  Michael> with locking mechanism headaches when a user with, say, an
  Michael> SMP Irix NFS client and an HP/UX NFS server has funny build
  Michael> problems.

I have been pondering this problem.  It's a shame Autoconf just can't
be fixed (and patches even backported to 2.13, if necessary) to change
the structure of the cache.  These locking problems exist because the
cache values are kept in a single file and to control access via a
single lock would be too coarse grained.

If, instead, the cache were a "dot" directory that contained files
whose filenames had the form "key=value", then test results could be
examined or updated using atomic file system operations and there
should be no races.

  Michael> So I'm in favor of a dumb, simple, and reliable mechanism
  Michael> to serialize the configures.

I think the performance of large configure runs is becoming too poor
to take this line for much longer.

Ben