On Mon, Aug 3, 2015 at 4:34 PM, Ganesh Ajjanagadde <gajja...@mit.edu> wrote: > On Sat, Aug 1, 2015 at 6:54 AM, Nicolas George <geo...@nsup.org> wrote: >> Le quartidi 14 thermidor, an CCXXIII, Michael Niedermayer a écrit : >>> > ffmpeg -lavfi testsrc -f framecrc | head >>> Thats interresting, i tried this and it did not trash my terminal >>> also ive almost never seen ffmpeg trash my terminal with the current >>> code. No matter how i run it >> >> As Ganesh noticed, I forgot the output "-" in my example, but you probably >> noticed that on your own, and that is not the issue. >> >> I just ran some tests and it appears that bash restores the tty state if the >> command was killed by a signal but not if it exited normally. >> >> Therefore, "ffmpeg ... - | head" will have the tty restored because ffmpeg >> dies by SIGPIPE, but "strace ffmpeg ... - | head" will trash the tty state >> because bash does not know of the SIGPIPE. The same goes for anything that >> runs ffmpeg as a sub-process, from "sh -c ..." to the FATE harness. >> >> Well, bash does part of its job, but I consider it should do more: leaving >> the tty trashed just because you inserted strace in front of the command is >> bad policy. > > Above example shows that even without this patch, tty state can be trashed. > Thus tty trashing is unavoidable, and only fix for this is via shell > configuration (see patch 2/3). > Therefore this patch should be included to fix Ticket2964.
Trashing the state even more often just because there is one example where this can already happen is not a valid reasoning. - Hendrik _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
