[PATCH] [ASan/Win] Use clang rather than clang-cl by default for lit tests. Make Windows-only tests explicitly use clang-cl.

[Timur Iskhodzhanov]

Do you think it's OK to add COMPILER_RT_TEST_CFLAGS which might be empty on
Linux if you prefer (or maybe -fno-rtti -g*?) but non-empty on Windows?


2014-05-28 23:13 GMT+04:00 Alexey Samsonov <vonosmas at gmail.com>:

>
> On Wed, May 28, 2014 at 2:06 AM, Timur Iskhodzhanov <timurrrr at google.com>
> wrote:
>
>> 2014-05-28 12:49 GMT+04:00 Timur Iskhodzhanov <timurrrr at google.com>:
>> > I should probably move the stuff we conditionally append to
>> SANITIZER_TEST_CFLAGS_COMMON to COMPILER_RT_GTEST_CFLAGS instead.  Will do
>> that in a separate commit.
>>
>> On the second thought, the _GTEST_ part of the macro name doesn't fit.
>
> Any objections if I define COMPILER_RT_TEST_CFLAGS in
>> cmake/Modules/AddCompilerRT.cmake ?
>> It should probably include other common stuff like -O2 and
>> -g[line-tables-only] ?
>>
>
> I'd prefer to keep GTest-relevant defines in _GTEST_ a separate variable
> as it's done at the moment. Not sure adding one more variable for unittest
> common compile flags would be good,
> e.g. I don't see why we should build all the tests with -O2 (one moment we
> may decide to test other opt levels). Listing these flags every time we add
> a specific unittest binary is not that bad.
>
>
> --
> Alexey Samsonov
> vonosmas at gmail.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140528/9e18b42a/attachment.html>