bug#10866: 1.11a OSX with llvm - 3 fails, 47 skips, 3 errors, out of 2856 tests

[Stefano Lattarini]

On 03/03/2012 04:57 AM, P. Martin wrote:
> 
> On Mar 2, 2012, Stefano Lattarini <address@hidden> wrote: 
>
>> Is perhaps your file system a case-insensitive HFS+?  If yes, does
>> the attached patch fix the failure?
>>
>> Thanks,
>>   Stefano
> 
> 
> Yes it's case sensitive,
>
Did you mean "insensitive" here, right?

> and your patch fixes the failure for java-uninstall.test.
> 
> My filesystem is the default type: Mac OS Extended (Journaled).
>
Thanks for confirming.  I've thus pushed the patch.

> I fixed my test setup.  I went back to using the system bison-2.3.
> I stopped setting MAKEINFO and HELP2MAN.
>
Note that your setup wasn't truly broken -- the automake testsuite should
ideally be able to deal with exported MAKEINFO and HELP2MAN transparently.

> The latter is in /usr/local/bin anyway.  And as if by the magic of
> Leap day...
> 
> No fails now from 1.11a, ( 57d0988 )
>
Good!  But ...

> # TOTAL: 2910
> # PASS:  2847
> # SKIP:  21
> # XFAIL: 41
> # FAIL:  0
> # XPASS: 0
> # ERROR: 1
>
... there's still this annoying error:

> ERROR: get-sysconf
> ==================
>
> Running from installcheck: no
> Using TAP: no
> PATH = 
> /private/tmp/homebrew-automake-HEAD-DUdI/tests/ax:/private/tmp/homebrew-automake-HEAD-DUdI/tests/wrap:/usr/local/Cellar/libffi/3.0.10/bin:/usr/local/Cellar/texinfo/4.13a/bin:/usr/local/Cellar/gettext/0.18.1.1/bin:/usr/local/bin:/usr/local/sbin:/usr/local/share/python:/usr/local/texlive/2011/bin/x86_64-darwin:/usr/bin:/bin:/usr/sbin:/sbin:/usr/X11/bin:/usr/local/Library/Contributions/examples
> ++ pwd
> /private/tmp/homebrew-automake-HEAD-DUdI/tests/get-sysconf.dir
> + st=0
> + test -d /private/tmp/homebrew-automake-HEAD-DUdI/.git
> + awk '
>    BEGIN { first = 1 }
>    (first == 1) { print; first = 0; next; }
>    /^[^\t]/ { exit(0); }
>    { print }
>  ' /private/tmp/homebrew-automake-HEAD-DUdI/ChangeLog
> awk: cmd. line:2: fatal: cannot open file 
> `/private/tmp/homebrew-automake-HEAD-DUdI/ChangeLog' for reading (No such 
> file or directory)
> + st=1
>
This is probably due to HomeBrew setting up the srcdir and builddir in a
way unexpected to get-sysconf.test.  Do you happen to know how one can
reliably determine whether the build system is being run by HomeBrew, so
that the failing part of the test can be skipped instead of poisoning
the whole testsuite result?

Thanks,
  Stefano