On Sat, 2012-10-20 at 11:04 +0200, Reinhard Tartler wrote: > notice them in our tests. This revised patch takes out the definition of > ff_sqrt_tab from avcodec without breaking the binaries. Obviously, we > get a small performance hit in the shared library case, but considering > the circumstances, I think that this is acceptable in this case.
Did the existing code really avoid the performance hit in libavcodec? As I posted about ff_log2_tab two days ago, at least in the case of that symbol none of the versions avoided the unnecessary overhead, because Libav still lacks compile-time visibility information. I'd expect this case to be the same. _______________________________________________ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel