Re: Odd Guile compilation problems

[Mark Kettenis <kettenis@wins.uva.nl>]

   From: Craig Brozefsky <craig@red-bean.com>
   Date: 01 Jun 1999 04:52:09 -0700

   >   Second: for some reason, the Hurd build produces a shared library named
   > 'libguile.so.4.0'.  I believe that this is out of whack with the expected
   > result of 'libguile.so.4.0.0' -- at any rate, debian/rules assumes at one
   > point that this file exists and that 'libguile.so.4' (also missing) exists.
   > Is this a problem in autoconf, automake, or libtool?  I notice that libglib
   > (which is also built with these programs) correctly produces the file
   > libglib-1.2.so.0.0.3 when built in the Hurd, so I'm hesitant to
   > implicate them.

   Hmm, this is interesting.  I see in another response to your message
   that it is a possible libtool issue.  Could someone give me a
   suggestion as to how the rules file should handle this?  is there some
   way I can force libtool to use a consistent versioning scheme?

What version of libtool is being used.  The current libtool (that is
1.3.2) looks broken to me for the Hurd.  Too little OS specific is
being supplied.  I'll investigate & fix.  The Hurd should behave very
similar to Linux.  All that would be needed for guile would be
upgrading to the fixed libtool :-).

Mark