- From: Edward Welbourne <eddy@chaos.org.uk>
- Date: Tue, 31 Dec 2002 20:28:25 +0000
- To: www-patentpolicy-comment@w3.org
The robots seem to have mangled their attempt to get me to ack that I'm happy to have an earlier mail archived ... so here's a quick summary of the principal point, just in case the mess isn't sorted out in time to get the original in by the deadline ... http://www.w3.org/TR/2002/WD-patent-policy-20021114/ Generally good, but ... Section 3, clause 3: [a W3C Royalty-Free license] > 3. may be limited to implementations of the Recommendation, and to > what is required by the Recommendation; What constitutes an implementation of the Recommendation ? If I implement some W3C Recommendations by writing a bunch of shared libraries along with some thin clients, no one of my shared libraries is an implementation of any one Recommendation. Yet this is a natural architecture for a developer to use, especially when several programs in the suite, or several instances of one of them, may be running at the same time on an O/S capable of saving resources by sharing the libraries between processes. If such a shared library implements some Essential Claim, it makes this functionality available to other applications. Indeed, someone implementing an application which has nothing to do with any W3C Recommendation could arrange to link against my shared library so as to access the patented functionality. If they're careful, they can deliver their application with a stub library, with the same API as mine but returning `sorry not implemented' responses to attempts to exercise the Essential Claim; the application links against this by default and copes with such failures if they arise; but works better if it gets linked at run-time against my library. This certainly can be done: and the application developer can fairly safely argue that they're not infringing the patent. It'd be silly (though possibly legally sound) to argue that the user who invoked the application is the patent infringer, even if wholly ignorant of the magic that is run-time linking. However, the patent-holder can certainly argue that my shared library is not an implementation of the relevant Recommendation(s), hence that it wasn't covered by the RF license. Thus it would appear that 3.3 will let a patent-holder insist that I not implement any Essential Claim in any shared library. Indeed, the whole point of 3.3 is clearly to let them insist that exemption from their patent not extend to applications other than those implementing W3C Recommendations: which can only be achieved if the Essential Claim isn't implemented via a shared library. This would be a drastic hindrance to application developers. Eddy.
Received on Tuesday, 7 January 2003 16:01:53 UTC