autoconf-patches
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: fyi-ac-arg-var.patch


From: Akim Demaille
Subject: Re: fyi-ac-arg-var.patch
Date: 21 Jun 2001 13:14:33 +0200
User-agent: Gnus/5.0808 (Gnus v5.8.8) XEmacs/21.4 (Academic Rigor)

>>>>> "Pavel" == Pavel Roskin <address@hidden> writes:

Pavel> In fact, the `A' in AU_DEFUN stands for Akim :-)

:)

Pavel> I believe that we should satisfy the following requirements:

Pavel> 1) AC_SYS_RESTARTABLE_SYSCALLS should work as before, i.e. it
Pavel> should expand to the same code in configure. As a consequence
Pavel> ...

The current implementation is OK, and AU it too.

Pavel> 2) it should still be possible to use AC_REQUIRE from
Pavel> AC_SYS_RESTARTABLE_SYSCALLS (it's used now).

The current implementation is OK, and AU it too.

Pavel> 3) Use of AC_SYS_RESTARTABLE_SYSCALLS in configure.in should
Pavel> cause a warning displayed by autoconf (that was the idea of the
Pavel> patch).

Likewise.

Pavel> 4) The testsuite should test
Pavel> AC_SYS_RESTARTABLE_SYSCALLS. Warnings from autoconf should be
Pavel> either ignored or checked for being non-empty.

If AU'd, that will be the case.

Pavel> 5) autoupdate should either leave AC_SYS_RESTARTABLE_SYSCALLS
Pavel> unchanged or add a comment saying that it's obsolete (the later
Pavel> would have ugly consequences if autoupdate is run more than
Pavel> once). Maybe it should tell the user that there is no
Pavel> replacement for AC_SYS_RESTARTABLE_SYSCALLS, and eliminating it
Pavel> requires manual intervention.

Hm...

Pavel> The existing testsuite implies that either a macro causes no
Pavel> warnings or it's updateable.

Not exactly.  Either the macro is tested `stupidly', or it has
dedicated code.  But what matters is that AC_DEFUN macros must not
depend upon obsolete macros.

Pavel> The simpleast solution would be to introduce AC_DEFUN_WARNING
Pavel> that would act like like AC_DEFUN except that mktests.sh would
Pavel> put it to a separate list of macros tested with, say
Pavel> AT_CHECK_MACRO_WARN, that would make sure that a warning is
Pavel> produced by autoconf.

I'm not too happy with adding some more forms of AC_DEFUN, and just
for the test suite :(  I'd rather have an exception for these macros,
and have dedicated code somewhere, passing the right -W flags to have
them silent.

Pavel> Having done that, we could improve autoupdate to print
Pavel> macro-specific warnings.

That's easy to do as is:

AU_DEFUN([FOO], [m4_warn([alksjdlkasj])[$0]])
AC_DEFUN([FOO], [asldk;sad])


Well, if you think AC_DEFUN_WARNING is the way to go, go.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]