[Ffmpeg-devel] overall license review - adding proper license headers

Rich Felker dalias
Fri Sep 1 17:39:03 CEST 2006


On Fri, Sep 01, 2006 at 04:13:47PM +0200, Diego Biurrun wrote:
> libavcodec/fdctref.c claims all rights reserved.  This is non-free.
> Houston, we have a problem.  Where does this file come from?

It's the reference implementation. IIRC it's not used and just
included for people to read but I may be mistaken.

> In libavformat/ dv1394.h is marked as GPL while dv1394.c is marked as
> LGPL.  While this is probably an oversight, this makes FFmpeg versions
> compiled with dv1394 support effectively GPL.  It's compiled in by
> default on platforms that support it.  This looks like a simple
> oversight, but probably the original author should be asked before this
> is changed or the code put under CONFIG_GPL.

Yes this probably won't be an issue but should be resolved.

> libavcodec/armv4l/simple_idct_arm.S is marked as GPL, but the original
> commit message from Alex says:
> 
>   Optimized simple idct for arm by Frederic 'dilb' Boulay <...>.
>   Currently licensed under the GPLv2, but the author
>   allowed to license it under the LGPL, feel free to change
> 
> I think this should simply be switched to LGPL then.

Yes.

> libavcodec/i386/idct_mmx.c is marked as GPL but with the exception that
> it can be used as LGPL in libavcodec.  IMO this is equivalent to
> releasing it under the LGPL in libavcodec.  The licensing note should be
> updated.

Yes, there is not such thing as "GPL for one particular project" or
"LGPL for one particular project" since both these licenses allow
unlimited relicensing under themselves. However this sort of language
from an author is very dangerous IMO because it indicates that they do
not understand GPL/LGPL at all and might try to attack users who
exercise their rights under the LGPL at a later time, claiming that
they did not license under the LGPL when in fact they did.

> libavcodec/ps2/idct_mmi.c seems to be from Intel Application Note
> AP-922, all the links in the file are 404, thus the license of this file
> is unclear.  Hmmm, not good.

Indeed. web.archive.org?

> libavcodec/liba52/resample_mmx.c is informally marked as "under GPL", I
> suggest adding the official GPL header instead.
> Same for libavutil/softfloat.h but with LGPL.

I hope we can remove liba52 at some point soon..

> libswscale/rgb2rgb.c and libswscale/rgb2rgb_template.c have a note
> 
>   palette & yuv & runtime cpu stuff by Michael (michaelni at ...) (under GPL)
> 
> but the rest was apparently written by Nick Kurshev and Alex.  I think
> the rest can be assumed to be GPL as well, IMO we should make this
> explicit with the official GPL header.

Agree.

> libavutil/adler32.c says
> 
>  * For conditions of distribution and use, see copyright notice in zlib.h
> 
> we don't include zlib.h so the conditions should be pasted there IMO.

Is adler32.c still used even?

> libavcodec/truemotion1data.h is taken from GPL code and relicensed under
> the assumption that data tables are not copyrightable.  Hmmmmm.

IMO this is totally valid. If you want, use gnu indent to remove all
information from the file aside from the actual data values.. ;)

> In libavcodec/ jfdctfst.c, jfdctint.c and jrevdct.c refer to a README
> file for conditions of distribution and use.  Where do they come from?
> What does that README file say?

Probably libjpeg.

> libavcodec/mpegaudiotab.h says that "The licence of this code is
> contained in file LICENCE found in the same archive".  What archive?
> What LICENSE file?

:)

> The files listed below have no license header.  I would suggest adding
> the proper/official (L)GPL license header to them to make things crystal
> clear.  Some are very short header files, but we have 5 line headers
> that do come with an LGPL header.  Some are data tables, but we do have
> data tables that come with an LGPL header.

Data tables with any copyright attached just make you look like a fool
IMO. See ESR's comments in terminfo.src for a good laugh. :)

> Thoughts/comments/flames welcome.

:)))

Rich





More information about the ffmpeg-devel mailing list