gettext-0.10.35 _nl_msg_cat_cntr bug

[Andreas Ley]

Hi!

There seems to be a glitch in gettext-0.10.35 when checking _nl_msg_cat_cntr.
Here's an excerpt from m4/gettext.m4, lines 57 to 75:

           if test "$gt_cv_func_gettext_libc" = "yes" \
              || test "$gt_cv_func_gettext_libintl" = "yes"; then
              AC_DEFINE(HAVE_GETTEXT)
              AM_PATH_PROG_WITH_TEST(MSGFMT, msgfmt,
                [test -z "`$ac_dir/$ac_word -h 2>&1 | grep 'dv '`"], no)dnl
              if test "$MSGFMT" != "no"; then
                AC_CHECK_FUNCS(dcgettext)
                AC_PATH_PROG(GMSGFMT, gmsgfmt, $MSGFMT)
                AM_PATH_PROG_WITH_TEST(XGETTEXT, xgettext,
                  [test -z "`$ac_dir/$ac_word -h 2>&1 | grep '(HELP)'`"], :)
                AC_TRY_LINK(, [extern int _nl_msg_cat_cntr;
                               return _nl_msg_cat_cntr],
                  [CATOBJEXT=.gmo
                   DATADIRNAME=share],
                  [CATOBJEXT=.mo
                   DATADIRNAME=lib])
                INSTOBJEXT=.mo
              fi
            fi

This is called when gettext has been found in libc or libintl, but the check
for _nl_msg_cat_cntr neglects to add -lintl in the latter case, thus installing
po files in $prefix/share in the first and $prefix/lib in the latter case.

Since this only has an effect if one installs a package on two platforms
that expose both cases (here: linux and aix) _and_ uses a common $prefix/share
but separate $prefix/lib's, this could easily be overseen.

Bye, Andy

-------------------------------------------------------------------------------
   Andreas Ley       !  "Even when you're   !  Email: address@hidden
   Goethestr. 27     !   a genius, life is  !  Home:  +49 721 84 10 36
   76135 Karlsruhe   !   a mystery!"        !  RZ:    +49 721 608 6341 <- NEU!
   Germany           !  Doogie Howser, M.D. !  AKK:   +49 721 96 403 22