www-jcp-open mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roy T. Fielding" <field...@gbiv.com>
Subject Re: [VOTE] New ASF/JCP Policies
Date Thu, 12 Jul 2007 10:37:47 GMT
On Jul 11, 2007, at 10:40 PM, Dain Sundstrom wrote:

> I think we are talking by each other.  I think this boils down to  
> one simple issue.
>   Is the intellectual property only licensed to compliant  
> implementations of the specification?
> If so, I don't see how this fits with the ASF.  FWIU, anyone should  
> be able to use, modify and redistribute derivatives of ASF  
> software.  Specifically, someone should be able to modify Geronimo  
> (JEE5 tested and certified) to make it non-compliant and still have  
> the ability to sell such a non-compliant derivative work.  Of  
> course they could not call it JEE or Geronimo, but they should be  
> able to sell their binary.
> Assuming all derivative works must be compliant with the spec in  
> order to receive the core IP rights contained in the software, then  
> I don't see how it qualifies for sections 3 and section 7 of the  
> OSI open source definition (http://www.opensource.org/docs/osd).

Nothing in intellectual property law is that simple.  To distribute
the software as open source, we need to have the right to distribute
under copyright law and not be aware of any incompatible restriction
under patent or trademark law.  The OSD assumes that the distributor
knows all possible restrictions on the software they are distributing,
which is absurd in practice.  [Actually, the OSD is a rather
self-serving document that includes a lot of stuff that has nothing
to do with the source code being open, but rather that enables
redistributors to package and sell it without paying royalties.]

Companies don't tell others what patents may apply to what software
until they accuse them of infringement, and even then the issue may not
be determined until the end of several years of litigation and a few
million bucks down the drain.  There would be no "open source" software
if the mere potential for patent claims was included in the definition.

The good news is that Apache figured all that out a long time ago.
We develop our own software under CLAs that are more than just
copyright licenses.  Likewise, the JCP has the reciprocal funnel
from EG members for compliant implementations that, if we make
the brave (and completely unprecedented) assumption that US patent
exhaustion rules might apply to software that has been licensed
once by virtue of having passed the TCK and make sure there are no
FOU restrictions on the patent license, might be enough to make
the source code be licensed by the EG for all necessary patents
even if they don't contribute to us directly under CLA and even
if the source code is later modified for derivative works.

All of that is enough for Apache to distribute its code as open
source, at least until someone tells us differently.


View raw message