[cfe-dev] Fix for preprocessor CommentHandler bug

Andy Gibbs andyg1001 at hotmail.co.uk
Sun Jun 17 07:40:34 PDT 2012


Hi,

Could I just bring this patch back into the light?

Here is the link to the original patch:
http://lists.cs.uiuc.edu/pipermail/cfe-dev/attachments/20120612/2b7041e4/attachment.obj

Are there any objections to getting this committed?

Thanks,
Andy


On Tuesday, June 12, 2012 7:07 PM, Andy Gibbs wrote:
> On Tuesday, June 12, 2012 6:31 PM, David Blaikie wrote:
>> <clippy>It looks like you're trying to fix a bug - would you like to
>> include a test case?</clippy>
>>
>> & is this a fix for: http://llvm.org/bugs/show_bug.cgi?id=13065 by
>> chance? I'm curious what caused the bug as I'd assumed it was due to
>> my commit ( https://llvm.org/svn/llvm-project/cfe/trunk@158093 ). But
>> judging by the fix, that might not be the case.
>
> No, the bug was uncovered while making an enhancement patch for the
> "-verify" feature, where I changed from the use of the separate parse step
> to using the CommentHandler interface.  Unfortunately, the CommentHandler
> interface was not invoked in some cases where comments followed #else or
> #endif directives, and this meant I needed to alter some pre-existing
> test-cases to enable them to work with my patch.  This fix was directed at
> removing this necessity.
>
> In reply, then, to the missing test-cases, I have test-cases as part of my
> patch for the new "-verify" feature, since this actually uses the
> CommentHandler interface and can directly test it.  I accept that this is 
> a
> weak reply, but my hope is that with the approval of this and my other
> recent preprocessor patch, then my "-verify" patch can also meet with
> approval -- in which case the test-cases will appear as part of that.
>
> Alternatively, is there a better way to test the CommentHandler interface?
> If so, I can supply some specific test-cases for this patch.
>
> Thanks
> Andy 




More information about the cfe-dev mailing list