On 19/01/12 15:21, Justin Ruggles wrote: > Only if an encoder has both CODEC_CAP_VARIABLE_FRAME_SIZE and > CODEC_CAP_DELAY set and doesn't use encode2(). We don't have any such > encoders, and any new audio encoders should use encode2(). > > The problem would be that nb_samples could be zero at that point, which > would mean we try to allocate a 0-size output packet. So an > av_assert0(nb_samples) would work too. It's really more of a "this > shouldn't ever happen" case anyway.
That reminds me that I'd like to avoid triggerable asserts and long ago I provided a specific AVERROR to signal such conditions. All in all the patch as is doesn't look right. lu -- Luca Barbato Gentoo/linux http://dev.gentoo.org/~lu_zero _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
