http://bugs.winehq.org/show_bug.cgi?id=33649
--- Comment #2 from Alexandre Rostovtsev tetromino@gentoo.org 2013-05-25 23:19:36 CDT --- Created attachment 44559 --> http://bugs.winehq.org/attachment.cgi?id=44559 alternative patch
I think the intent of AC_LANG_PROGRAM here was to test for linking validity by making sure that a pointer to the gnutls_mac_get_key_size() function would be usable at runtime. The patch in comment #1 replaces this test with something fundamentally different: a test for a type definition that matters only at compile time.
I therefore think the patch I am attaching would be a better solution.