2017-02-12 5:36 GMT+01:00 Reindl Harald <[email protected]>: > > Am 12.02.2017 um 03:30 schrieb Carl Eugen Hoyos: >> >> 2017-02-12 3:23 GMT+01:00 Reindl Harald <[email protected]>: >> >>> gcc-6.3.1-1.fc24.x86_64 >> >> gcc.gnu.org doesn't know this release;-) >> >> Anyway: Please understand that several hundred real, user-reported, >> reproducible bugs exist in current FFmpeg, some of them regressions. >> I hope you agree that compiler bugs are not our priority... >> (It's bad enough that we have to invest time in "undefined behaviour" >> where no compiler ever failed.) > > no i do *not* agree beause some some of that warnings may the > reason for unfixed bugs
I wonder what "may" means here... > things like "src/libavcodec/mpegvideo.c:960:17: warning: this > 'if' clause does not guard... [-Wmisleading-indentation]" are a clear > indication that this code should be reviewed Please do;-) > and to be frankly where i develop software (though not in C/C++ but that > don't matter) i first keep my house clean and fix any warnings software can Sounds as if we found a new contributor - welcome! > detected automatically and *then* consider to waste my time trying to > reprodce things which could have been gone by get rid of all the code smell It's funny that you wrote this mail a few hours after I fixed most of the current warnings - or "several" if you prefer. (Yes, we do exactly what you ask for, I just request that you realize fixing these warnings practically never fixes a real issue, while we know about many - ! - real issues including remote DOS, regressions on valid input files and sometimes subtle change of behaviour, both intended and unintended.) >> That's apart from the fact that --extra-cflags and --extra-ldflags >> generally make bug reports invalid (as you correctly indicated) >> except for paths and libraries > > you still refused to answer my question if you did a LTO build as you (You refused to answer my question.) > pretended you can't reporduce it, as you always request full input > of ffmpeg command lines why don't you do the same? I was hoping my original comment made it crystal-clear that I did try to reproduce the issues you saw. > src/libavformat/rtpdec_formats.h:41:5: warning: type of > 'ff_h264_handle_aggregated_packet' does not match original declaration > [-Wlto-type-mismatch] > int ff_h264_handle_aggregated_packet(AVFormatContext *ctx, PayloadContext > *data, AVPacket *pkt, > ^ > src/libavformat/rtpdec_h264.c:206:5: note: > 'ff_h264_handle_aggregated_packet' was previously declared here > int ff_h264_handle_aggregated_packet(AVFormatContext *ctx, PayloadContext > *data, AVPacket *pkt, > ^ > src/libavformat/rtpdec_h264.c:206:5: note: code may be misoptimized unless > -fno-strict-aliasing is used So what is your initial analysis of this warning? (You can find mine above, the line starts with "I hope".) Carl Eugen _______________________________________________ ffmpeg-user mailing list [email protected] http://ffmpeg.org/mailman/listinfo/ffmpeg-user To unsubscribe, visit link above, or email [email protected] with subject "unsubscribe".
