On Sat, Sep 10, 2016 at 11:07:29AM +0800, Alex Vong wrote: > From 066bee5d7795c6caf2376be4a38e2cf0b397ef98 Mon Sep 17 00:00:00 2001 > From: Alex Vong <alexvong1...@gmail.com> > Date: Sat, 10 Sep 2016 00:32:03 +0800 > Subject: [PATCH 2/3] gnu: vlc: Add configure flags. > > * gnu/packages/video.scm (vlc)[arguments]: Add configure flags to enable > most fast-math optimizations. > --- > gnu/packages/video.scm | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/gnu/packages/video.scm b/gnu/packages/video.scm > index 858e007..a041666 100644 > --- a/gnu/packages/video.scm > +++ b/gnu/packages/video.scm > @@ -650,7 +650,10 @@ audio/video codec library.") > `(;; Gross workaround for > <https://trac.videolan.org/vlc/ticket/16907>. > ;; In our case, this led to a test failure: > ;; test_libvlc_equalizer: libvlc/equalizer.c:122: test_equalizer: > Assertion `isnan(libvlc_audio_equalizer_get_amp_at_index (equalizer, > u_bands))' failed. > + ;; FIXME: This bug is fixed upstream. For now, we enable: > "ac_cv_c_fast_math=no" > + "CFLAGS=-ffast-math -fno-finite-math-only" > + "CXXFLAGS=-ffast-math -fno-finite-math-only"
I don't know much about this subject. Can anyone say if we can expect this to work on all our supported architectures?
