[cfe-dev] [PATCH] Fix clang test failures on MIPS

[Manuel Klimek]

On Wed, May 9, 2012 at 12:52 PM, Simon Atanasyan <satanasyan at mips.com> wrote:
> On Wed, May 9, 2012 at 11:37 AM, Manuel Klimek <klimek at google.com> wrote:
>> While Chandler seems to have understood the idea, I'm still slightly
>> confused in what the test would look like - perhaps I need to see
>> it... :)
>>
>> Also, I think this would test that the mechanism for setting adjusters
>> works - in my opinion it's more important to have a smoke test for the
>> syntax-only default case first (which will not ensure that we're able
>> to change the adjuster, but that the default adjuster works the way we
>> want it to).
>>
>> That is, while I think more specific tests are valuable and adding
>> them will make me happy, I'd try to come up with a test that checks
>> the current fsyntax-only behavior first.
>
> The attached patch has a comment about the default setting for the
> adjuster in the ClangTool and new test case Tooling/multi-jobs.cpp.
>
> If we pass -no-integrated-as option to the clang, it always starts two
> jobs. The first one produces assembler file. The second one runs
> assembler to generate an object file. Tooling library cannot handle
> two jobs and the test is failed on all platforms. When the patch is
> applied ClangTool uses ClangSyntaxOnlyAdjuster to add -fsyntax-only
> option to the command line. With that option clang does not start
> assembler and the test is passed.

Ah, that makes sense :)
One small nit: can you model the test after this one:
http://llvm.org/viewvc/llvm-project/cfe/trunk/test/Tooling/clang-check-args.cpp?revision=154989&view=markup

I don't think it's necessary for that test to check the json parsing layer...

Apart from that LGTM.

Thanks!
/Manuel