On Tue, Apr 24, 2012 at 11:04 PM, David Blaikie <[email protected]> wrote:
> On Tue, Apr 24, 2012 at 1:54 PM, Nico Weber <[email protected]> wrote: > > Hi, > > > > the attached patch expands the include_next for <float.h> on windows > > to happen for _msc_ver too. > > > > The test case for this is: > > > > #include <float.h> > > > > inline bool IsFinite(const double& number) { > > return _finite(number) != 0; > > } > > > > But since that test depends on having MinGW's or MSVC's float.h, I > > didn't add it to the patch. > > Could you just restrict the test to certain platforms? (I've not > looked at how fine-grained our ability to do that with lit is, but you > could do it with similar macros in the test file than those in the > header... but that's perhaps a little redundant) > > - David > > > > > Ok? > > > > Nico > > > I know of two mechanism in lit: - REQUIRES (such as REQUIRES: shell) - XFAIL (such as XFAIL: mingw, win32) I do not know if REQUIRES could be used with platforms (as XFAIL) and blacklisting does not seem very interesting... Is there a set of builtin such as __has_include that would detect whether an include_next exist ? -- Matthieu
_______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
