autoconf
[Top][All Lists]
Advanced

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

Re: autoconf-2.65 has 252 test failures on OpenSolaris x86


From: Ralf Wildenhues
Subject: Re: autoconf-2.65 has 252 test failures on OpenSolaris x86
Date: Mon, 30 Nov 2009 21:32:41 +0100
User-agent: Mutt/1.5.20 (2009-08-09)

* Dr. David Kirkby wrote on Mon, Nov 30, 2009 at 11:19:28AM CET:
> Ralf Wildenhues wrote:
> >* Dr. David Kirkby wrote on Sun, Nov 29, 2009 at 11:46:52PM CET:
> >>configure: warning: line 946: " quote may be missing
> >>configure: warning: line 955: `...` obsolete, use $(...)
> >>configure: warning: line 960: `...` obsolete, use $(...)
> >>configure: warning: line 962: " quote may be missing
> >>configure: warning: line 993: `...` obsolete, use $(...)
> >>configure: warning: line 1014: `...` obsolete, use $(...)
> >>configure: warning: line 1036: `...` obsolete, use $(...)
> >>configure: warning: line 1041: $ not preceeded by \

> >It's pretty ironic that a Solaris shell should warn about `...`, with
> >Autoconf using that construct mostly because shells like Solaris <=10
> >10) /bin/sh does not cope with $(...).  Oh well.  Can the shell be told
> >to not output these warnings somehow?  I took a quick look at
> ><http://dlc.sun.com/osol/man/downloads/current/> but couldn't find a
> >switch or environment variable to do so.

> Yes, it is ironic. Does this mean autoconf is likely to be usable,
> or will the fact it issues these warnings during the tests mean
> autoconf will not work?

I haven't found any failures in your log file that weren't caused by
this verbosity, so yes, autoconf is likely to be usable, but the
configure scripts will produce quite some clutter output on this system.

The patch below should avoid the bulk of the failures (merely making the
testsuite less strict; I don't see why such a patch should make it into
the Autoconf repository); there are a number of configure invocations
that don't use AT_CHECK_CONFIGURE that would still fail with this.

I don't think that the verbosity itself will cause many configure
scripts to produce wrong results, if at all; tests checking stderr
contents from command substitutions are rare and fragile.

> I've asked on comp.unix.solaris.

Thanks.

Cheers,
Ralf

diff --git a/tests/local.at b/tests/local.at
index ed7e509..a977e26 100644
--- a/tests/local.at
+++ b/tests/local.at
@@ -233,7 +233,7 @@ m4_define([AT_CHECK_CONFIGURE],
 [AT_CAPTURE_FILE([config.log])[]dnl
  AT_CHECK([./configure $1],
          [$2],
-         m4_default([$3], [ignore]), [$4])])
+         m4_default([$3], [ignore]), m4_default([$4], [ignore]))])
 
 
 # AT_CHECK_ENV




reply via email to

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