Hi Ronald, On Sat, Jul 8, 2017 at 2:33 PM, Ronald S. Bultje <rsbul...@gmail.com> wrote: > > I can see the design from the patch. > > What's missing is a justification for the downside of the design, which is > that updates to this variable by the user are no longer propagated to the > worker threads.
My justification is the YAGNI principle. Although the current code allows the FF_DEBUG_THREADS option to be toggled dynamically, I believe that was not intended, and I believe nobody actually does that. In my (admittedly limited) code search, I only see the FF_DEBUG_THREADS option set via the -debug thread_ops command-line option. > The syncing is extremely trivial (simply by moving the > assignment from init to update_from_user) and has no threading implications > (since it's a boolean, so you can make it atomic) so this really shouldn't > be a big deal to amend. Yes, the syncing is trivial to add. When someone actually needs to toggle the FF_DEBUG_THREADS option dynamically, we can easily add the syncing at that time. This is your call. Please let me know your decision. Thanks! Wan-Teh Chang _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
