[cfe-commits] r157890 - /cfe/trunk/unittests/Tooling/ToolingTest.cpp

012/6/3 Hal Finkel <hfinkel@anl.gov>:

Author: chapuni
Date: Sat Jun 2 10:34:26 2012
New Revision: 157890

URL: http://llvm.org/viewvc/llvm-project?rev=157890&view=rev
Log:
Revert r157819, "#ifdef out a broken test on win32"

FYI, LLVM_ON_WIN32 is useless.

Can you be more specific? I thought that configure defined these (and I
assumed that the other build systems did too as appropriate). Is that
not true?

LLVM_ON_WIN32 requires "config.h". In this case, it went useless lack
of config.h.

IMO, we can assume obtaining condition to rely on _WIN32, not
LLVM_ON_WIN32, on the source tree.
LLVM_ON_WIN32 might be useful out of C/C++ sources (eg. scripts), I think.

Should we honor LLVM_ON_WIN32 yet?

...Takumi

012/6/3 Hal Finkel <hfinkel@anl.gov>:
>> Author: chapuni
>> Date: Sat Jun 2 10:34:26 2012
>> New Revision: 157890
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=157890&view=rev
>> Log:
>> Revert r157819, "#ifdef out a broken test on win32"
>>
>> FYI, LLVM_ON_WIN32 is useless.
>
> Can you be more specific? I thought that configure defined these
> (and I assumed that the other build systems did too as
> appropriate). Is that not true?

LLVM_ON_WIN32 requires "config.h". In this case, it went useless lack
of config.h.

Are you using cmake?

-Hal

Both, autoconf and cmake on mingw msys. :wink:
I don't imagine it would be an issue whether I am using cmake.
Let me know if you know any issues.

...Takumi