[PATCH xorg-gtest] aclocal: fill AS_IF with noop statements

Chase Douglas chase.douglas at canonical.com
Wed Aug 8 09:20:07 PDT 2012


On 08/06/2012 05:33 PM, Peter Hutterer wrote:
> On Mon, Aug 06, 2012 at 09:59:29AM -0700, Chase Douglas wrote:
>> On 07/29/2012 06:05 PM, Peter Hutterer wrote:
>>> Expanding this macro fails on some machines (RHEL6). It expands to
>>> if test <condition>; then
>>>
>>> fi
>>>
>>> Empty statements are invalid.
>>>
>>> Signed-off-by: Peter Hutterer <peter.hutterer at who-t.net>
>>> ---
>>>   aclocal/xorg-gtest.m4 |    2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/aclocal/xorg-gtest.m4 b/aclocal/xorg-gtest.m4
>>> index 6efff08..0f6aa85 100644
>>> --- a/aclocal/xorg-gtest.m4
>>> +++ b/aclocal/xorg-gtest.m4
>>> @@ -47,7 +47,7 @@ AC_DEFUN([_CHECK_GTEST],
>>>
>>>     AS_IF([test "x$have_gtest" = xyes],
>>>           [AC_SUBST(GTEST_CPPFLAGS)]
>>> -        [AC_SUBST(GTEST_SOURCE)])
>>> +        [AC_SUBST(GTEST_SOURCE)] [:])
>>>   ]) # _CHECK_GTEST
>>>
>>>   # CHECK_XORG_GTEST([ACTION-IF-FOUND], [ACTION-IF-NOT-FOUND])
>>
>> This seems like a bug in the RHEL6 autotools. When I run autoreconf
>> here, almost all the conditional statements end up looking like:
>>
>> if test <condition>; then :
>>      <statements>
>> fi
>>
>> If there are no statements, the ':' takes care of things.
>>
>> A quick google search doesn't turn up any helpful information. Are
>> you in a position to try to figure out why this is broken on RHEL6?
>> If it's too much trouble, we can apply this patch. However, I think
>> it really is an issue with autotools.
>
> looks like this was fixed in autoconf upstream with v2.63b-16-g52b83b8
> (and a follow-up bc79643). We ship 2.63* in RHEL6, so this patch isn't in
> yet. We only require 2.60 in xorg-gtest so we should ship this fix (I'm
> obviously not in favour of bumping the autoconf requirement past RHEL6 :)
>
> Cheers,
>    Peter
>
> * note that apparently 2.63b is equivalent to what we would tag 2.63.99.901,
>    i.e 2.63b is newer than 2.63

Thanks for figuring this out! I'm ok with having a work around where 
necessary, so this is fine by me:

Reviewed-by: Chase Douglas <chase.douglas at canonical.com>

I then edited the commit message to be more descriptive about the issue 
(autoconf versions, etc.) and bumped the serial number at the top of 
xorg-gtest.m4.

This has been pushed as commit 714341ae8136ac958ec555694a23dadf80b1c852.

Thanks!

-- Chase


More information about the xorg-devel mailing list