> Can anybody suggest a strategy for isolating the problem? I can't easily
> remove much code because the servers won't get anything they can use if
> I do.
Someone needs to look at the garbage collection of threads. I suspect
that there is a bug in the thread GC code. I can do that, but not
until next week.
> 1) .x dependencies not in yet; aclocal ... fixed.
There is no bug in the distribution itself. The bug is in the script
generating snapshots. Only Jim or Pat can fix this.
> 2) in scm-sigs.c usleep is asked to return an int. On my 2.0.30 linux box
> it is void
usleep returns an int on Sun Solaris machines. We should add
something to the configuration script to handle this. I'll try to fix
it soon.
> 3) dynamic loading unhappy because scmconfig.h defines has HAVE_LIBDL
> while dynl.c looks for HAVE_DLOPEN.
This is again the snapshot script.
You can fix points 1 and 3 yourselves by installing automake-1.3 and
configuring with --enable-maintainer-mode. (You may have to remove
some files lingering from earlier snapshots.)
/mdj