On Wed, Jun 26, 2013 at 9:01 PM, Derek Buitenhuis <[email protected]> wrote: > On 2013-06-26 2:57 PM, Hendrik Leppkes wrote: >> Am i the only one that thinks that this patch is extremely ugly, and >> if a compiler generates such broken code, it should perhaps simply not >> be used on non-Intel CPUs? > > I don't mind, if it's opt-in. It's not much more ugly than how we override > vsnprintf for MSVC. Being able to debug in MSVC's debugegr with inline ASM > enabled seems like a benefit to me. YMMV. We could feasibly segrated it in > compat/intel_cpu.{c|asm}, as Luca suggested. >
But inline asm can't even be enabled yet, and the patches to allow that are also quite ugly. :) Personally i don't miss inline asm when debugging. The only benefit would be a faster library, but i wouldn't trust ICL enough to provide a reliable build, of course thats just IMHO. Right now the YASM code is always build, so it would be better if all this uglyness is not even touched at all by other compilers. _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
