[PATCH] Move duplicate Windows-specific compiler flags to a common CMake variable

[Reid Kleckner]

Are all the warnings in gtest headers or in the Windows SDK or CRT?  There
are lots of pragmas we don't understand in windows.h, for example.  We
should already understand all of gtest, but maybe I'm wrong.


On Thu, May 29, 2014 at 1:35 PM, Timur Iskhodzhanov <timurrrr at google.com>
wrote:

> A-ha, so you suggest to use -isystem rather than -I for gtest, right?
> The system headers should be fine out of the box already?
> (Please note I don't use clang-cl there now)
>
> 2014-05-30 0:26 GMT+04:00 Reid Kleckner <rnk at google.com>:
>
> Clang suppresses all warnings in system headers, which are put onto the
>> include search path with the -isystem flag.  In clang-cl, we consider
>> anything in %INCLUDE% to be a system header by passing it with -isystem to
>> clang -cc1.
>>
>>
>> On Thu, May 29, 2014 at 11:41 AM, Timur Iskhodzhanov <timurrrr at google.com
>> > wrote:
>>
>>> Can you elaborate on the -isystem approach? I'm not sure I know what
>>> you're taking about.
>>> 29 мая 2014 г. 21:03 пользователь "Reid Kleckner" <rnk at google.com>
>>> написал:
>>>
>>> drive by nits that were already there
>>>>
>>>> ================
>>>> Comment at: cmake/Modules/AddCompilerRT.cmake:129
>>>> @@ -126,1 +128,3 @@
>>>>  if(MSVC)
>>>> +  # MSVC system headers and gtest use a lot of deprecated stuff.
>>>> +  list(APPEND COMPILER_RT_TEST_CFLAGS
>>>> ----------------
>>>> Can we use -isystem to suppress these?  Does -D_CRT_SECURE_NO_WARNINGS
>>>> solve this problem?
>>>>
>>>> ================
>>>> Comment at: cmake/Modules/AddCompilerRT.cmake:137
>>>> @@ +136,3 @@
>>>> +
>>>> +  # We should teach clang to understand more pragmas.
>>>> +  list(APPEND COMPILER_RT_TEST_CFLAGS
>>>> ----------------
>>>> ditto, this can be handled with -isystem.
>>>>
>>>> http://reviews.llvm.org/D3952
>>>>
>>>>
>>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140529/8f146c9c/attachment.html>