On Fri, 27 Jan 2017 22:24:35 +0200 (EET)
Martin Storsjö <[email protected]> wrote:

> > Than we have to postpone opening any file handle until the first call to
> > dash_flush, right?
> > Which should be possible, but requires some code to be moved. Using a
> > dyn_buf (patch 10) might also help here, so writing to files happens as
> > late a possible.  
> 
> Hmm, indeed. If we postpone this patch until after that, this should be 
> easier to handle right. (The alternative would be to just count the 
> bitrate of the codec payload as we pass into the chained muxer; the 
> bit_rate value we use otherwise also refers to only the codec payload data 
> not including the container, but since we're probably doing the dyn_buf 
> stuff anyway, that's even simpler.)

Mh, after thinking about this a bit more, I'm not sure the complexity is worth
it.

I'd have to store the contents of the dyn_buf containing the init data somewhere
until the first packet is ready to be flushed so I can calculate the average
bitrate to be used in the filenames. Only then I know the init segment filename.

This might be ok for mp4, since everything happens within the first call to
dash_flush. But for webm this doesn't work that easy.

So maybe the patch is ok for now.

Regards
Peter

Attachment: pgpOGQYlx22fg.pgp
Description: OpenPGP digital signature

_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to