[FFmpeg-devel] [PATCH v3 0/5] Support for stream dispositions in MP4

[Jan Ekström]

On Mon, Sep 20, 2021 at 6:00 PM Jan Ekström <jeebjp at gmail.com> wrote:
>
> Compared to v2:
> * aviobuf changes to make a function useful in MP4 null-delimited string
>   parsing into AVBPrint.
> ** Extended read_line_to_bprint to be a more generic read_string_to_bprint.
> ** Added a maximum length argument to read_string_to_bprint.
> ** Added a new function ff_read_string_to_bprint_overwrite.
>

I would like to get more eyes on this part of the patch set, since the
C char string stuff is in my own opinion the bit where I have the
largest possibility of messing it up.

> * Switched from buffers with hard-coded sizes in avformat/mov to AVBPrint based
>   parsing with ff_read_string_to_bprint_overwrite.
>
> * Minor changes to the actual primary changes of the patch set:
> ** avformat/isom: "KindWritingModeCMAF | KindWritingModeUnifiedOrigin" now
>                   has spaces between the bitmasks for better readability.
> ** avformat/{isom,movenc}: add and utilize KindWritingModeNB for the maximum
>                            value limit in the AVOption definition.
>
> First patch implements the CMAF specified way of flagging what in FFmpeg
> are are called stream dispositions. Other identifiers such as HTML media track
> kinds are allowed, but if there is a DASH identifier for something, it should
> be utilized in stead.
>
> Second patch is a compatibility patch for one of the vendors that supports this
> feature. If this is considered a too bad of a hack, we can drop it from being
> upstreamed, but at least I wanted to bring it up :) . The compatibility mode
> is not the default, so it should also not proliferate such behavior.

If there are no further comments, I will start pulling this in
tomorrow or so, with a micro version bump to note the addition of this
feature (stream dispositions) within an existing feature (movenc), as
it is API-wise not an addition of new API end points.

Jan