Re: AC_DECL_SYS_SIGLIST is broken in latest autoconf

[Mike Castle]

In article <address@hidden>,
Ben Pfaff  <address@hidden> wrote:
>address@hidden (Mike Castle) writes:
>> But then it's not the same as AC_CHECK_DECL([sys_siglist]), 
>
>Why should it be the same as AC_CHECK_DECL([sys_siglist])?  It is
>specialized for sys_siglist, so it should do what is necessary.

Then include a patch to update the docs as well?

>> and it's listed as an obsolete macro anyway.
>
>So what?  If you're going to support an obsolete macro at all
>then why not support it right?

To discourage people from using it?

It looks like this macro was the only one simplified.  Others may be broken
as well.

>> >--- autoconf-2.57.orig/lib/autoconf/specific.m4
>> >+++ autoconf-2.57/lib/autoconf/specific.m4
>> >@@ -60,7 +60,12 @@
>> > # -------------------
>> > AN_IDENTIFIER([sys_siglist],     [AC_CHECK_DECLS([sys_siglist])])
>> > AU_DEFUN([AC_DECL_SYS_SIGLIST],
>> >-[AC_CHECK_DECLS([sys_siglist])
>> >+[AC_CHECK_DECLS([sys_siglist],,,[
>> >+#include <signal.h>
>> 
>> I wrap the #include in #ifdef HAVE_SIGNAL_H myself.
>
>Sounds like a good idea.

Hmmm.  2.13 also included sys/types.h.  Is that necessary on any systems?
(If so, need to update my usage).

Odd.  coreutils just uses plain sys_siglist with no special headers.
Wonder if it actually works.

Time to catch the train now though.

Cheers,
mrc
-- 
     Mike Castle      address@hidden      www.netcom.com/~dalgoda/
    We are all of us living in the shadow of Manhattan.  -- Watchmen
fatal ("You are in a maze of twisty compiler features, all different"); -- gcc