Zhao Zhili (HE12025-04-11): > Background: > > 1. There is a av_uninit macro which suppress uninitialized variable warning > > #if (defined(__GNUC__) || defined(__clang__)) && !defined(__INTEL_COMPILER) > # define av_uninit(x) x=x > #else > # define av_uninit(x) x > #endif
Yes. > 2. Declaration after statement coding style is allowed now I fail to see the connection. Oh, reading further, I guess you think moving the declaration of the variable will be enough to let the compiler know the variable is not used and silence the warning. I strongly doubt it might have that effect. > 3. There is a crash fixed by ab792634197 which is related to av_uninit, and I > have confronted it again recently So, what does it prove? It proves that 16 years ago (e658657528d) Diego silenced a warning without making sure the logic itself was sound. Which is exactly what you are proposing to do right now in a different way. > 1. Deprecated av_uninit and removed it in the future > > 2. Remove usage in current code Not before or without 3. > 3. When there is uninitiated variable warning > > a. Declare variable when need if it’s possible, so it can be initiated at the > same time > b. Use valid default value if there is > c. If there is no valid default value, or you think the variable never being > used uninitialized, assign an insane value and add assert before use if > > av_assert1(value != invalid_value); Are you volunteering to do that huge work? Running FATE with valgrind and no optimization seems like a quicker way of finding the real bugs. Regards, -- Nicolas George _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
