This is the mail archive of the
automake@gnu.org
mailing list for the automake project.
Re: Running a plain ./config.status at top level
- From: Alexandre Duret-Lutz <duret_g at lrde dot epita dot fr>
- To: Automake List <automake at gnu dot org>
- Date: Sun, 11 May 2003 22:31:48 +0200
- Subject: Re: Running a plain ./config.status at top level
- References: <mv465ppgeow.fsf@nostromo.lrde.epita.fr><2003-04-09-20-34-25+32318+duret_g@lrde.epita.fr><1106.82.64.23.203.1049914381.squirrel@www.lrde.epita.fr><2003-04-10-00-05-50+2954+duret_g@lrde.epita.fr><mv4vfwwqz4j.fsf@nostromo.lrde.epita.fr><20030430132717.A22688@telepres.com>
>>> "Eric" == Eric Siegerman <erics@telepres.com> writes:
[...]
adl> $? works at least on GNU, Solaris, *BSD, OSF/Tru64, and
adl> HP-UX. (It's POSIX, of course.)
Eric> Solaris 7's make(1) says:
>> o make attempts to derive values for the dynamic macros
>> `$*', `$<', and `$?', while processing explicit tar-
>> gets. It uses the same method as for implicit rules;
>> in some cases this can lead either to unexpected
>> values, or to an empty value being assigned. (Actu-
>> ally, this was true for earlier versions as well, even
>> though the documentation stated otherwise.)
I'm not really worried by this. I don't think unexpected applies
to $?.
[...]
Eric> But then, why not do the whole thing the "make" way, using "::"
Eric> rules, rather than adding yet more embedded shell-scripting?
This sounds really attractive, but I'm more concerned because
::-rules are not POSIX. Does someone know a Make implementation
that doesn't support them? I've made a note to add a check for
::-rules to the Automake test suite, so we get reports if they
don't work somewhere.
--
Alexandre Duret-Lutz