On 5/8/17 8:44 AM, wm4 wrote: > This is mostly for convenience, so that API users (including possibly > libavcodec) don't have to duplicate native format mapping tables. > > Not sure if these should really have a device_ctx as argument. hw_format > is probably enough, but I wasn't sure whether future implementations > might possibly depend on runtime state. The intention is not to return > _supported_ formats only, so I don't foresee any actual need for a > device_ctx, but it might also be short-sighted to require one.
Probably it is safer to keep the argument and maybe it is better to not take it as AVBufferRef but ask for the unwrapped pointer directly. > There is also the question what to do if multiple hwcontext > implementations support the same hw_format. Should (hw_format, > sw_format) be enough to uniquely map to a native format? You might want to just return the first or alternatively you could populate an array, but maybe it is overkill. lu _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
