[Top][All Lists]

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

Re: calling autoreconf and minimizing rebuilds

From: Stepan Kasal
Subject: Re: calling autoreconf and minimizing rebuilds
Date: Thu, 7 Sep 2006 16:20:19 +0200
User-agent: Mutt/

Hello David and Ralf,

I also did some experiments, and the results are similar to what Ralf

Indeed, when I did
        automake -a
then automake changed the autom4te.cache even though it should not.
It seems it refreshed the output.1 and traces.1, though the result
was the same as in the previous run of autoconf.

(I have not observed a new traces.x and output.x file, the last
ones (traces.1 and output.1 in my case) were touched.)

When I compared the state of autom4te.cache before and after the
automake run, I saw this:
--- autom4te.sav/requests       2006-09-07 15:27:28.000000000 +0200
+++ autom4te.cache/requests     2006-09-07 15:27:42.000000000 +0200
@@ -355,2 +355,3 @@
                         'AC_TYPE_PID_T' => 1,
+                        'AM_ENABLE_MULTILIB' => 1,
                         'AC_CONFIG_FILES' => 1,

So the fix is obvious: add the macro to the preselect list in
autom4te.  I installed autoconf-20060907-presel.patch (attached)
to the Autoconf CVS.

Well, if this mistake happens again, it might help if autoreconf
ran autoconf after automake.  But I decided against this, because:

1) autoheader has to be ran before automake, to generate

2) autoconf should be ran before autoheader, because the comment in
autreconf says it has better diagnostics.  (I hope it is still true.)

3) Calling autoconf twice (before and after automake) is too evil to
Bruno, who has autom4te caching switched off.

Well, we could do one thing:
we should add a test to automake test suite to verify that
`automake -a' after `autoconf' does not touch autom4te.cache/output.1
(No I do not volunteer right now, sorry.)


Attachment: autoconf-20060907-presel.patch
Description: Text document

reply via email to

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