okay, here's the relevant section from libtool, just a few
lines before the prog-wrapper generation:
# Only actually do things if our run command is non-null.
if test -z "$run"; then
# win32 will think the script is a binary if it has
# a .exe suffix, so we strip it off here.
case $output in
*.exe) output=`echo $output|sed 's,.exe$,,'` ;;
esac
# test for cygwin because mv fails w/o .exe extensions
case $host in
*cygwin*) exeext=.exe ;;
*) exeext= ;;
esac
$rm $output
trap "$rm $output; exit 1" 1 2 15
okay, could someone please enlight me what that is about?
>From what I see, on $build=anywin it shall not be called
.exe for, er, some kind of reason. Is that still needed in
the current libtool version? And even if it is, would it be
allowed to ifdef this section with a $build!=anywin test?
My proposal - enclose both of the above case-statements
with a test for win-$build-compile, so I can have a
shell-.exe on unix-systems atleast - and someone with
experience in win-$build systems, please check if the above
happens to be actually junk-code that should better be
dumped right away.
however, how do we detect a win-$build system in libtool?
well, $build is not set, so at the moment I use a patch
that uses a test -z "$exeext". Is that var set on a
win-$build system? 'dunno....
what do you think? TIA, guido
--- ltmain.sh.1.4 Wed Apr 25 01:35:10 2001
+++ ltmain.sh Sat Apr 28 01:00:08 2001
@@ -3512,16 +3512,18 @@
# Only actually do things if our run command is non-null.
if test -z "$run"; then
# win32 will think the script is a binary if it has
# a .exe suffix, so we strip it off here.
+ if test -n "$exeext" ; then
case $output in
*.exe) output=`echo $output|sed 's,.exe$,,'` ;;
esac
# test for cygwin because mv fails w/o .exe extensions
case $host in
*cygwin*) exeext=.exe ;;
*) exeext= ;;
esac
+ fi
$rm $output
trap "$rm $output; exit 1" 1 2 15
$echo > $output "\
_______________________________________________
Libtool mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/libtool
