[FFmpeg-devel] [PATCH] avformat: Implement subtitle charenc guessing

wm4 nfxjfg at googlemail.com
Sat Dec 13 16:12:55 CET 2014

On Sat, 13 Dec 2014 12:34:28 +0100
Nicolas George <george at nsup.org> wrote:

> First, your patch seems to happen after the text demuxers have parsed the
> text files. Therefore, this can not work for non-ASCII-compatible encodings,
> such as UTF-16. You might say that UTF-16 already works, but its
> implementation is bogus and leads to user-visible problems (see trac ticket
> #4059). But even if it was not, we would not want two competing detection
> layers.

For someone who complains about "bickering and lack of enthusiasm" you
sure bicker and discourage a lot.

What about it is bogus? #4059 is a problem with ffmpeg.c and the stuff
in lavc, not the general approach. In fact, the UTF-16 change made
UTF-16 just work with any API user.

Recoding in the demuxer is unacceptable, because it makes it impossible
to change the codepage later or get any kind of user interaction. Doing
it on file opening unnecessarily complicates these things.
Detection can't be done in lavc (and the way lavc does recoding, with
fatal errors on failure, is also plain unacceptable).

I agree with the comment though that ffmpeg isn't going to be linked to
all these fancy detection mechanisms. This is mostly because you have
to enable external libraries explicitly when compiling, so usually they
won't be picked up.

More information about the ffmpeg-devel mailing list