Jan Drabner <jan@...> writes: > But I never asked for support. I reported what is in my > opinion bug.
Let's see. > swr_convert crashes under some circumstances, Definitely. > without telling any reason. Yes and no. If you had provided a backtrace as requested on http://ffmpeg.org/bugreports.html I suspect one would have been able to see what's wrong. (Everybody was able to guess what's wrong but it is often a bad idea to rely on guessing when answering questions here.) The relevant functions take a pointer to input data structures and a pointer to output data structures as parameters. In your case, you initialised the conversion functions saying "I want to convert to a planar format", so the actual conversion function expected you to pass a pointer to pointers. But instead you passed a pointer to allocated memory (it may have been different but this is what generally happens). Do you really think there is a possibility for the conversion function to know (in advance) that dereferencing the pointer you passed will lead to a segfault? And do you agree that it is impossible to "tell a reason" in this case? > If that is not a bug, or a severely lacking feature or > documentation, I don't know what is. It may be missing documentation (it possibly is) but then you should not write "crash" but instead send a documentation update (now) to ffmpeg-devel. > I still have no idea why it should not be possible > to convert from any format to any other. That is exactly what it does. > sws_scale can do that just fine. But it will of course also (immediately) crash if you pass a pointer to RGB24 after initialising for GBRP. Carl Eugen _______________________________________________ Libav-user mailing list [email protected] http://ffmpeg.org/mailman/listinfo/libav-user
