On 2022/03/04 10:39, Klemens Nanni wrote: > > > > We could also disable the flag that triggers this? > > Meh, that only "worked" because my sparc64 tree was still at snappy > 1.1.8 which builds anyway... > > > > ok? > > OK kn for your patch
Should be all ports-gcc rather than just sparc64. Maybe use !defined(__clang__) ? > > > > > > --Kurt > > > > > > Index: patches/patch-snappy-stubs-internal_h > > > =================================================================== > > > RCS file: patches/patch-snappy-stubs-internal_h > > > diff -N patches/patch-snappy-stubs-internal_h > > > --- /dev/null 1 Jan 1970 00:00:00 -0000 > > > +++ patches/patch-snappy-stubs-internal_h 3 Mar 2022 05:28:19 -0000 > > > @@ -0,0 +1,14 @@ > > > +Don't set this for sparc64, it keeps it from building > > > + > > > +Index: snappy-stubs-internal.h > > > +--- snappy-stubs-internal.h.orig > > > ++++ snappy-stubs-internal.h > > > +@@ -99,7 +99,7 @@ > > > + #endif > > > + > > > + // Inlining hints. > > > +-#ifdef HAVE_ATTRIBUTE_ALWAYS_INLINE > > > ++#if defined(HAVE_ATTRIBUTE_ALWAYS_INLINE) && !defined(__sparc64__) > > > + #define SNAPPY_ATTRIBUTE_ALWAYS_INLINE __attribute__((always_inline)) > > > + #else > > > + #define SNAPPY_ATTRIBUTE_ALWAYS_INLINE > > > > > >
