[cfe-commits] r133196 - in /cfe/trunk: include/clang/Basic/DiagnosticSemaKinds.td lib/Sema/SemaExpr.cpp test/SemaCXX/null_in_arithmetic_ops.cpp test/SemaCXX/nullptr_in_arithmetic_ops.cpp test/SemaObjCXX/null_objc_pointer.mm

[Chandler Carruth]

On Fri, Jun 17, 2011 at 1:53 PM, Chad Rosier <mcrosier at apple.com> wrote:

> Thanks for taking care of this.
>

I just turned this warning back on with r133384. I've checked and we now
have tests in the Clang suite that should cover the same cases as the GCC
test suite covered. I'll revert if I see smooshy fail on it though and
investigate.


>
>  Regards,
>   Chad
>
> On Jun 17, 2011, at 1:48 PM, Richard Trieu wrote:
>
> Chad,
>
> Revision 133287 puts the new warning behind a flag and set to default
> ignore, which should fix your test case (I think).  I'll continue working on
> the warning to iron out the problems.
>
> Richard.
>
> On Thu, Jun 16, 2011 at 7:05 PM, Chad Rosier <mcrosier at apple.com> wrote:
>
>>
>> On Jun 16, 2011, at 6:34 PM, Richard Trieu wrote:
>>
>> Hey Chad,
>>
>> I am unfamiliar with how these cases are set up.  What are the flags that
>> are used to run this and what is the expected behavior from this test?  It
>> looks like the dg-warning marks which lines are allowed to have warnings,
>> and not others.  Is this correct?
>>
>>
>> As Chris pointed out, I would see
>> http://llvm.org/svn/llvm-project/clang-tests/trunk/gcc-4_2-testsuite/README.txtfor directions on running the test suite.  I believe your assumption about
>> dg-warning is correct, but I'm by no means an expert of how the nightly
>> testers work.
>>
>> I think the problem might be because my patch introduces a warning at
>> lines 32 & 33, which aren't expecting a warning.  One is a comparison of
>> NULL and a function.  The other is a comparison of an array and NULL.  Is
>> this warning appropriate here or should these cases be excluded from the
>> warning?
>>
>>
>> Seems like a reasonable guess.  Unfortunately, I don't know the intent of
>> your patch or the expected output from the individual test case, so I can't
>> really voice an opinion on the matter.  Once you get the test suite running
>> locally you could try modifying the test case by adding the appropriate
>> dg-warning "" markers.  If this fixes things and you are confident the
>> emitted error message is correct then you could submit a patch to the test
>> case to make the buildbot happy.
>>
>>  Chad
>>
>>
>> Richard.
>>
>> On Thu, Jun 16, 2011 at 5:44 PM, Chad Rosier <mcrosier at apple.com> wrote:
>>
>>> Sorry about that!  I see how that could make debugging difficult.
>>>
>>> The reported error is:
>>> g++.old-deja/g++.other/null1.C (test for excess errors)
>>>
>>>  Chad
>>>
>>> On Jun 16, 2011, at 5:36 PM, Chandler Carruth wrote:
>>>
>>> Please don't post links to the internal Apple build bots, post the text
>>> of the regression? =] Then we can make progress on which of the several test
>>> cases in that file is misbehaving.
>>>
>>> On Thu, Jun 16, 2011 at 5:32 PM, Chad Rosier <mcrosier at apple.com> wrote:
>>>
>>>> See:
>>>>
>>>> http://smooshlab.apple.com:8013/builders/gccTestSuite_clang-x86_64-darwin10-RA__c%2B%2B/builds/854
>>>>
>>>
>>>
>>>
>>
>>
>
>
> _______________________________________________
> cfe-commits mailing list
> cfe-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20110619/d2304347/attachment.html>