License Approval Process

John Cowan jcowan at
Wed Feb 16 17:20:29 UTC 2000

"Dennis E. Hamilton" wrote:

> I think I understand how this works.

You do.

> AA.  The Angels right to make their own derivative works of X is diminished
> to the extent that the Angels do not have an automatic license to take
> additions and modifications from the derivative works produced by others.

I wouldn't use the word "diminished"; there is *never* any such "automatic
license" to reuse other people's works.  Only an explicit license can
make that possible.

> In doing so, there needs to be care to
> avoid infringing the intellectual property rights of Borg, Inc. and the
> Cavaliers.

A clean-room approach is essentially always sufficient, and may be
more than is required.  (The FSF uses such an approach w.r.t. Unix
source code; if you've seen the AT&T source, you can't work on the GNU version.)

> Let's have "M admits-derivative N" represent that license N is automatically
> admissible on derivatives of works for which license M is automatically
> available.
> Then  MIT admits-derivative GPL
>         MIT admits-derivative MIT
>         MIT admits-derivative closed
>         GPL admits-derivative x if-and-only-if x = GPL
          MPL admits-derivative MPL
          MPL admits-derivative closed

A good summary, to which I have added the MPL conditions
To extend the summary to the LGPL, you have to have two relations,
because LGPLed works admit two kinds of derivatives -- informally speaking,
the "extension of library" type and the "use of library" type.
LGPL works like GPL for the former, and like MIT for the latter.
Trying to distinguish between these in an architecture-neutral way
is what makes the LGPL so lengthy.

Schlingt dreifach einen Kreis vom dies! || John Cowan <jcowan at>
Schliesst euer Aug vor heiliger Schau,  ||
Denn er genoss vom Honig-Tau,           ||
Und trank die Milch vom Paradies.            -- Coleridge (tr. Politzer)

More information about the License-discuss mailing list