RE: Apache License : grant of patent license and derivative works

[Jeffrey Thompson]

"William A. Rowe, Jr." <[EMAIL PROTECTED]> wrote on 03/31/2005 01:10:19 PM:

> I see that the ASL clause doesn't include derivative language;
>
It doesn't cover all possible derivative works.  As was pointed out earlier, open source licenses (with one notable exception that I'm aware of) don't provide patent grants that cover all possible derivative works.

Open source licenses that don't have an explicit patent grant in them (such as the modified BSD license) rely on an implied patent license.  While the exact scope of the implied license isn't certain, it would likely only cover only those uses of the code specifically intended -- probably only the project to which it was directly contributed.  The current ASL language provides much better patent coverage.  It covers not only the original code in the original project, but it covers all derivatives of that code in any ASL licensed project.  Theoretically, one could ask for a patent grant which was unlimited, but no major project does that.

The current ASL license provides greater rights in this respect than the GPL, EPL, BSD, SISSL, and many others I can't think of off the top of my head.

> Three things occur to me;
>
> *) We are missing the derivative clause for patents.  If we did add
Its missing only if it should be there and it isn't.

>    this in the ASL 2.1 or 3.0, contributors have agreed to be bound
>    to the language of future licenses.  But it seems that due to how
>    narrowly and specifically this clause was written, would a revised
>    clause 3. be sufficient to bind previous patent contributions?
The precise effect of a license upgrade provision is uncertain when the new license asks for broader rights than the original.  I'd be surprised, however, if that worked.

>
> *) Does the fact that this was narrowly written actually increase
>    the position of the patent holder, to the detriment of the user?
>    Were it better had we said nothing about patents, v.s. this
>    specific language?
I don't think so.  The implied patent licenses are likely not that broad.  If any of the other lawyers on the list have a case or two on point, that'd be appreciated, because I don't have one.

>
> *) Why did we limit the impact of the self-destruct clause within
>    the patent license to terminating patent licenses alone?  If we
>    spark GTW (global thermonuclear war, a running joke among members)
>    why would we explicitly preserve their copyright license grants?
My understanding of the intent of that limitation was specifically to avoid GTW.  If there is a patent claim, the defensive suspension is limited to the patent licenses.  People would still have their copyright licenses, with the options to modify the code if necessary to avoid the patents at issue (which is what we want if there is a blow-up about patents).  To pull the copyright licenses too seems to me to be much too onerous.  The user is faced with, not rewriting a few modules to avoid a patent, but replacing the entire ASF licensed project.  That might not even be possible.  A commercial entity, especially a smaller one, might not be able to survive that.  In the end, I guess it is how ruthless does the ASF want to be.  

Jeff

Staff Counsel, IBM Corporation  (914)766-1757  (tie)8-826  (fax) -8160
(notes) [EMAIL PROTECTED]  (internet) [EMAIL PROTECTED] (home) [EMAIL PROTECTED]
(web) http://www.beff.net/