Recent CVS build failures

[John Daschbach <jldasch at 3-cities dot com>]

Builds from CVS fail for me after a Jan 9 CVS update (ok) Jan 10 not
ok.  The failure seems to be due to a change in the dynamic linking
code, or really the inlclude files for this.   The error is:

Making all in libguile
make[1]: Entering directory `/home/guile/guile/guile-core/libguile'
./guile-doc-snarf dynl.c -DHAVE_CONFIG_H -I. -I. -I. -I.. -I./..   -g
-O2 -Wall -Wmissing-prototypes dynl.c > dynl.x \
|| { rm dynl.x; false; }
dynl.c:235: ltdl.h: No such file or directory
make[1]: *** [dynl.x] Error 1

ltdl.h is a libtool file, but it doesn't appear that the guile build is
using libtool.  I have ltdl.h and a recent libtool.

[guile@hydrogen guile-core]$ slocate ltdl.h
/usr/share/libtool/libltdl/ltdl.h
[guile@hydrogen guile-core]$ libtool --version
ltmain.sh (GNU libtool) 1.3.4 (1.385.2.196 1999/12/07 21:47:57)
[guile@hydrogen guile-core]$ 

I even wiped the guile directory clean and did a new cvs checkout,
configure, make cycle with no change.   What I have not figured out is
that one of my work machines runs a nearly identical linux installation
(RH 6.0) and compiles todays guile cvs sources with no trouble.

This isn't a guile problem exactly, but these kind of build problems are
annoying.   The key is only partially the change from Jan 9 to 10 (which
is in the ChangeLog), the question is why it builds at work and not at
home.  Is it that something in the configure run at home is chosing
dynamic linking and not at work?  (On the other hand I've used dynamic
loading in both SCM and Octave on the work machine so one version of it
works there).

Any help on where to start would be great.

-John