[FFmpeg-devel] [RFC] change to the "setfield" filter

[Stefano Sabatini]

On date Monday 2012-04-16 13:50:22 +0100, Tim Nicholson encoded:
> On 16/04/12 12:47, Nicolas George wrote:
> > L'octidi 28 germinal, an CCXX, Stefano Sabatini a écrit :
> >> Why not to add a new value for "progressive"? This way you preserve
> >> the old semantics and don't break scripts.
> > 
> 
> I was trying to keep to the commonly used -1|0|1 format for consistency
> with other options that use the same type and range of arguments, and
> since Stefano came up with the script relatively recently, in response
> to a query I made, I thought the risk of breaking scripts was relatively
> low.

-1/0/1 (disable/enable/auto) was based on other tri-state options we
have in ffmpeg and other filters, but apparently this is not a
compelling reason for keeping it here, since we don't have a
disable/enable/auto logic.

> > The more values we add, the more nonsensical their numeric code will become.
> > Better not to expose it at all. This is already done for -1/0/1, but this is
> > not a reason to do it for new values as well.
> > 
> > I do not propose to break scripts, I only propose that scripts that use
> > -1/0/1 get a new warning:
> > 
> 
> I think I was suggesting that my proposal might break scripts, not that
> your suggestion might.
> 
> > [setfield] Using -1/0/1 is deprecated, please use auto/tff/bff.
> > 
> > And new scripts that use the new features will, perforce, use the names.
> >
> 
> A good argument. Is this consistent with other areas?

Maybe no (MPlayer filters tended to use numeric non-positional
parameters, and we kept them for not changing syntax), but I think we
may apply it in this specific case.
-- 
FFmpeg = Fabulous and Frenzy Mysterious Power Enigmatic Glue