Re: (no subject)

[Tim Van Holder]

> | 2001-09-16  Tim Van Holder  <address@hidden>
> |     * lib/autoconf/functions.m4: (AC_FUNC_GETPGRP) Don't rely on
> |     setpgrp() being present.
> |     (AC_FUNC_LSTAT_FOLLOWS_SLASHED_SYMLINK): Use conf$$ instead of
> |     conftest for test files.  Clean up after DJGPP's 'ln -s' (which
> |     creates conf$$.exe).

So all that remains is this:

Index: functions.m4
===================================================================
RCS file: /cvs/autoconf/lib/autoconf/functions.m4,v
retrieving revision 1.46
diff -u -r1.46 functions.m4
--- functions.m4        2001/08/22 06:24:46     1.46
+++ functions.m4        2001/09/17 20:14:35
@@ -692,10 +692,12 @@
 [AC_CACHE_CHECK(
        [whether lstat dereferences a symlink specified with a trailing
slash],
        [ac_cv_func_lstat_dereferences_slashed_symlink],
-[rm -f conftest.sym conftest.file
-echo >conftest.file
-if ln -s conftest.file conftest.sym; then
-  AC_RUN_IFELSE([AC_LANG_PROGRAM([AC_INCLUDES_DEFAULT],
+[dnl No point in running this test if ln -s doesn't work.
+if test "$LN_S" = "ln -s"; then
+  rm -f conftest.sym conftsym.exe conftest.file
+  echo >conftest.file
+  if ln -s conftest.file conftest.sym; then
+    AC_RUN_IFELSE([AC_LANG_PROGRAM([AC_INCLUDES_DEFAULT],
     [struct stat sbuf;
      /* Linux will dereference the symlink and fail.
         That is better in the sense that it means we will not
@@ -704,12 +706,13 @@
                 [ac_cv_func_lstat_dereferences_slashed_symlink=yes],
                 [ac_cv_func_lstat_dereferences_slashed_symlink=no],
                 [ac_cv_func_lstat_dereferences_slashed_symlink=no])
-else
-  # If the `ln -s' command failed, then we probably don't even
-  # have an lstat function.
-  ac_cv_func_lstat_dereferences_slashed_symlink=no
+  else
+    # If the `ln -s' command failed, then we probably don't even
+    # have an lstat function.
+    ac_cv_func_lstat_dereferences_slashed_symlink=no
+  fi
+  rm -f conftest.sym conftest.file
 fi
-rm -f conftest.sym conftest.file
 ])

 test $ac_cv_func_lstat_dereferences_slashed_symlink = yes &&

Is that OK to commit?
I also noticed that _AC_FUNC_FORK contains an 8-bit character (ü/u:) in
a
C comment.  Are there still compilers/preprocessors around that might
choke
on that (I doubt it, but I thought I'd check)?