autoconf-patches
[Top][All Lists]
Advanced

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

Re: FW: [GNU Autoconf 2.65] testsuite: 35 239 271 272 273 274 275 276 27


From: Ralf Wildenhues
Subject: Re: FW: [GNU Autoconf 2.65] testsuite: 35 239 271 272 273 274 275 276 277 278 279 280 failed
Date: Sun, 28 Feb 2010 15:09:51 +0100
User-agent: Mutt/1.5.20 (2009-10-28)

* Ralf Wildenhues wrote on Sun, Feb 28, 2010 at 02:12:00PM CET:
> I can reproduce these two failures on FreeBSD 6.4 (i386):
> 
> > 35. tools.at:1174: testing ...
> > ../../tests/tools.at:1191: : > sub/some-file && exit 77
> > stderr:
> > /home/users/wilx/tmp/autoconf-2.65/objdir/tests/testsuite.dir/at-groups/35/test-source:
> >  cannot create sub/some-file: Permission denied
> > stdout:
> > ../../tests/tools.at:1191: exit code was 2, expected 1
> > 35. tools.at:1174: 35. autom4te cache creation (tools.at:1174): FAILED 
> > (tools.at:1191)
> 
> > 239. c.at:57: testing ...

> > checking whether the C compiler works... no
> > ../../tests/c.at:66: exit code was 1, expected 77
> 
> The second is us failing to adhere to '(autoconf.info)Shell Functions':
>      Likewise, the state of `$?' is not reliable when entering a shell
>   function.
> 
> in AS_ERROR.  The first seems to be a genuine shell error difference
> (from glancing at SUSv3 I can only gather that a failed redirection
> should result in an exit status > 0, but no further specification).
> 
> I'm assuming the rest of the failures, which happen because no Fortran
> compiler is installed, are due to the same issue as 239.

I'm testing these two patches right now.  I didn't see an easy way to
keep as_fn_error API-compatible in the presence of trailing optional
arguments, so aren't we glad we didn't publish that function.  Still
feel a bit uneasy about the first patch, as we don't cover AS_ERROR
fully in the testsuite (e.g., the two-argument version).  So, close
review is appreciated.  Thanks.  (Sending them with format-patch in
case you want to go ahead and apply them, Eric.)

The second patch needs a subshell because sh just errors out right away
upon failed redirection, which is actually correct as ':' is a special
built-in utility.

I'm seeing spurious testsuite failures when I pass '-v -d -x NNN' as
TESTSUITEFLAGS, NNN being numbers, on FreeBSD.  Haven't investigated
further yet.

Cheers,
Ralf

Attachment: 0001-Fix-AS_ERROR-for-FreeBSD-sh.patch
Description: Text document

Attachment: 0002-Fix-autom4te-cache-creation-testsuite-failure-on-Fre.patch
Description: Text document


reply via email to

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