On Wed, Jul 23, 2014 at 7:03 AM, Anton Khirnov <an...@khirnov.net> wrote: > > On Tue, 22 Jul 2014 09:38:42 +0200, j...@v2v.cc wrote: >> On 07/21/2014 09:35 PM, Anton Khirnov wrote: >> > >> > On Sun, 20 Jul 2014 15:28:36 +0200, j...@v2v.cc wrote: >> >> On 07/20/2014 03:13 PM, Anton Khirnov wrote: >> >>> Could you share the sample? >> >>> This really sounds like a bug elsewhere that should be fixed. >> >> >> >> Here a short clip http://v2v.cc/~j/samples/emtpy_language.mpg >> >> >> > >> > Then your patch does not seem to work. >> > This was the source of my initial confusion, since in the patch you're >> > testing >> > for a NULL pointer in the tag value, but the dict API should not allow >> > that to >> > happen ever. But in this sample, the value is not a NULL pointer, but an >> > empty >> > string (i.e. a valid pointer pointing to a zero byte), which is a perfectly >> > valid dictionary entry (though admittedly not a valid language >> > description). >> >> you are right, not sure why my test did not fail with the first patch. >> the second patch to mpegts.js works as expected though. >> >> > Perhaps we could use the table from libavformat/avlanguage.c to actually >> > check >> > whether the language is valid. >> >> that might be a good idea too, do you know if this kind of thing is done >> somehere else in libav already? > > Does not seem so. > > But adding a function for checking whether a given string is in the given > codespace in the table should be simple enough. >
(this should just provide some context for the previous two pings) -- Vittorio _______________________________________________ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel