Work even if aclocal is invoked by hand

Problem reported by Ulrich Müller (bug#79974).
* configure.ac (gl_ALIGNASOF): Arrange for it to be defined
even if aclocal is invoked other than via admin/merge-gnulib.
Also, change comments from # to dnl as they don’t make
much sense in the shell script.
This commit is contained in:
Paul Eggert 2025-12-09 18:03:10 -08:00
parent bc65fb7486
commit b9b0c7ed51

View file

@ -3277,12 +3277,20 @@ if test "$emacs_cv_struct_alignment" = yes; then
structure to an N-byte boundary.])
fi
# Check for alignas now rather than waiting for gl_INIT to do it,
# as the system_malloc=no test below needs $gl_cv_header_working_stdalign_h.
# Do not simply call gl_INIT here, as the ac_cv_func_pthread_sigmask=yes
# assignment below affects how gl_INIT works.
# Instead, call gl_ALIGNASOF, the only part of gl_INIT needed here.
# Then make gl_ALIGNASOF a no-op so gl_INIT does not duplicate its code.
dnl Work around a glitch if aclocal is invoked outside of autogen.sh.
dnl Such an invocation generates an aclocal.m4 that lacks m4/stdalign.m4.
dnl The following hack checks the otherwise-uninvolved macro gl_STDALIGN_H
dnl so that the hack itself does not generate a diagnostic.
m4_divert_push([KILL])
m4_ifdef([gl_STDALIGN_H], [], [m4_include([m4/stdalign.m4])])
m4_divert_pop([KILL])
dnl Check for alignas now rather than waiting for gl_INIT to do it,
dnl as the system_malloc=no test below needs $gl_cv_header_working_stdalign_h.
dnl Do not simply call gl_INIT here, as the ac_cv_func_pthread_sigmask=yes
dnl assignment below affects how gl_INIT works.
dnl Instead, call gl_ALIGNASOF, the only part of gl_INIT needed here.
dnl Then make gl_ALIGNASOF a no-op so gl_INIT does not duplicate its code.
gl_ALIGNASOF
AC_DEFUN([gl_ALIGNASOF])