[Ffmpeg-devel] Patch add function to read record time in dv video

[Roman Shaposhnick]

On Sun, Sep 04, 2005 at 09:18:06AM -0600, Fred Rothganger wrote:
> Michael Niedermayer wrote:
> 
> >patch looks ok, roman, do you agree with that? IIRC we had some
> >disagreements in the past about how to pass this info around
> > 
> >
> 
>    Not to be too self-critical, but I think this is a pretty hacky way 
> to do it.  A nice way might be to put metadata in a name-value list 
> attached to some appropriate structure, or perhaps have a standard 
> function for probing metadata.  For example:

  Absolutely agreed! Fred, can you try to pursue this approach ? I'd
  it'll be really good to have the infrastructure in place first, so
  that we can solve generic DV metadata passign issue once and for all.

  I've been meaning to do this for quite some time now, but I'm really
  swamped at the moment (though, anticipate a comeback if anyone cares ;-)).

  As for the original patch of yours, I hope it won't be a problem if we hold 
  the putback for a little while longer, to see how this can be solved in
  a generic way. Wouldn't you agree ?

> The semantics of the names would be defined by the individual format, 
> but there would be a lot of commonality between codecs.  The value would 
> be an arbitrary structure, but well-defined for the given name.  For 
> example, it could be a null terminated string or a time structure, etc.  
> "size" would be the size of the structure.
> 
> The function would return metadata in the following priority order:
>    * Relative to the most recently decoded frame
>    * Relative to the container as a whole
> 
> If there are multiple streams and we want to be stream specific, a 
> metadata function could be added to AVCodec as well, or perhaps stream 
> could be one of the parameters of the function.
> 
> This approach could replace some of the metadata currently hard-coded 
> into AVFormatContext, such as title, author, copyright, etc.  For 
> writing such values to a stream, a metadata function can be added to 
> AVOutputFormat.

  I'm much better at looking at the code, so would you care to submit a
  patch ? ;-)

Thanks,
Roman.