On 08.11.2017 14:30, Martin Storsjö wrote: > On Wed, 8 Nov 2017, Jacek Caban wrote: > >> Hi Martin, >> >> >> On 11/7/17 10:29 PM, Martin Storsjö wrote: >>> I've tested our support for ucrtbase.dll a bit further by trying >>> building >>> a handful of libraries with it, and fixing the issues I run into. With >>> these patches, I'm able to build among others Qt (tested with 5.7.1). >> >> That's good to hear. Great work! >> >>> Some libraries have an issue where autoconf tries to detect whether >>> e.g. >>> snprintf or similar exist, by just checking if linking to a function >>> with that name succeeds. Since there is no function with that name in >>> the import library, it will fail and think that a replacement needs to >>> be provided, which then clashes with the real ones in stdio.h. >>> (Configure >>> checks that actually include the headers generally work though.) >> >> A solution for that would be to have non-inline __stdio_common_* >> wrappers inside libucrtbase.a. > > Yup, I thought about that - that's probably a sensible thing to do. > Does that mean we should get rid of the inline versions in the header > altogether, or do they play well together? (I guess I'll try and find > out.)
Both solutions should work, I think. I can see some advantages in removing inlines to reduce #ifdef complexity of headers (those parts are already hard to follow), but I don't have a strong opinion. Jacek ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
