On Wed, Dec 04, 2013 at 03:28:10PM +0200, Martin Storsjö wrote: > On Tue, 3 Dec 2013, Martin Storsjö wrote: > >On Mon, 2 Dec 2013, Diego Biurrun wrote: > > > >>This is more robust and does not interfere with host libc separation. > >>--- > >>configure | 2 +- > >>1 file changed, 1 insertion(+), 1 deletion(-) > > > >Why does check_header interfere with host libc separation? Just > >because we doesn't want to add a version of that test, to limit > >the size of the final patch in the series? > > I guess the issue is that check_header does enable/disable flags. We > could do something similar to check_header without the flag side > effects by doing "check_cpp_condition _mingw.h 1" as well.
Yes. > That is - I don't see that big benefit in handling mingw32 and -64 > separately with respect to the libc tests right now, but perhaps it > might be useful for something later? My point mainly is that I'm not > too keen on complicating this when the current test can be kept in a > different form. > > OTOH the current mingw libc code here is pretty ugly actually and I > could agree that the new form (in the latest version of the patch) > is better (regardless of the host libc detection set). The latter is the goal here. Even though we currently add the same set of flags for both MinGW32 and MinGW64, this might not be the case forever. Besides the end result is cleaner IMO. I have two printf/stdio CPPFLAGS patches for mingw libc pending that could never be done properly w/o host and target libc splitting. I will revisit those patches once this set is in. Possibly (not necessarily) the flags needed for both mingw variants will already be different then. Diego _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
