[FFmpeg-devel] [RFC] drop libamr support

[Benjamin Larsson]

Diego Biurrun wrote:
> I've been thinking about libamr again while handling the SaffronEncoder
> and I am thinking more and more that we should drop libamr support
> outright.

Well I disagree. I think that we should just only support it via dynamic
loading. No need to remove support completely for libamr.

> 
> [Reminder: libamr is copyrighted without any kind of license grant.   As
> a consequence it is non-free and unredistributable.]
> 
> Whatever binaries are created with libamr support are unredistributable.
> Nonetheless many are created and redistributed and I fear that most
> people are not doing this on purpose but are blissfully unaware of the
> consequences.

Not if only dynamic loading is supported.

> 
> Another point is that while there is a convenient way to hook up libamr
> in FFmpeg the incentive to implement a native replacement will be low.
> 
> I fear we are not doing anybody a favor with what we currently have and
> that the disadvantages outweigh the advantages.
> 
> Anybody who desperately needs AMR support can go back to a revision with
> libamr support, people who wish to compare native replacements with the
> reference lib can do the same.

Well there is no encoder for amr so I think we should keep support for
libamr just for the sake of not loosing the amr encoder option.

> 
> I vote for removing libamr.
> 
> Diego

I vote for not removing libamr. What we should do is change the license
to undistributable when libamr is staticly compiled with FFmpeg in GPL mode.

MvH
Benjamin Larssonn