[Theora] Independent implementations?

[Christoph Lampert]

<Pine.LNX.4.58.0407132048120.23066 at login.math.uni-bonn.de><000501c46953$fb65eea0$60389c3f at computername>
<40F4F0B6.1080109 at gmx.net><002601c469bb$536511f0$33389c3f at computername>
<40F564E5.8010908 at gmx.net><000701c469c5$82a7b800$8c649c3f at computername>
<40F575B5.5030108 at gmx.net>
<000e01c469d0$54ce8070$40389c3f at computername>
<40F58932.4060309 at gmx.net>
<Pine.LNX.4.58.0407142253200.27310 at login.math.uni-bonn.de>
<40F5A6C7.6030605 at gmx.net>
Message-ID: <Pine.LNX.4.58.0407150853240.27907 at login.math.uni-bonn.de>

On Wed, 14 Jul 2004, Maik Merten wrote:
> "6.2. Effect of New Versions.
> Once Covered Code has been published under a particular version of the
> License, You may always continue to use it under the terms of that=20
> version. You may also choose to use such Covered Code under the terms of=
=20
> any subsequent version of the License published by On2. No one other=20
> than On2 has the right to modify the terms applicable to Covered Code=20
> created under this License."
>=20
> So you can choose.

You could choose, _if_ it was On2 who changed the license, not someone=20
else (because this other person wouldn't have the right to do so). So,=20
we still need a document about how the VP3-stuff in theora stuff got its=20
new license.=20

> Most people with programming skills don=B4t live in caves. It=B4s highly=
=20
> likely that any implementation compatible with Ogg Theora is written=20
> with the help of the Ogg Theora specification (which _is_ derived from=20
> VP3-code). Is any code written with the help of this specification=20
> automatically derived from the VP3-codebase (which surely is covered by=
=20
> the irrevocable patent license)? Then there would be no problem....

No no no, absolutely not. Please again be careful not to mix copyright and=
=20
patents. "derived from" means copyright, so it has to be based on the=20
sources.

With just a documents of the theora specs, you can never create derived=20
works of VP3 or Theora. No other software license applies to your code,=20
except of what you want it to.=20
Also, you can read any kind of code (even Microsofts "shared source") and=
=20
then implement the same thing in a different way, without creating=20
"derived works", or anything. It would be derived if you just=20
cut-and-paste it (this includes trivial changes, like change of variable=20
names), but not if you start from scratch, just with a description of what=
=20
to do.=20
It's _good_ that it is that way, because otherwise, nobody would ever be=20
allowed to write a second, independent implementation of anything, under=20
any different license than the original.=20

But, if the specs say, "for decoding as theora bitstream you must_ do=20
this and that", and if you then do that in your implementation (in your=20
own way), and then there is a patent claiming that nobody is allowed to do=
=20
"this and that" without a license (or rather, to export the product to the=
=20
US, if it contains a routine to do "this and that"), then you need a licens=
e,=20
no matter how you started. If you started with VP3 and created derived work=
s,=20
you have it automatically, because On2 gave it to anyone who did that.=20
If you didn't, you might not have it, at least, the question is still not=
=20
clear to me.=20

Christoph=20

P.S. Of course, IANAL.