MPL, CDDL, patents and licencing

[Jason White]

First, I apologize in advance if this post is off-topic. According to
the OSI Web site, the mailing list is for discussion of licences that
are under review, but a perusal of the archive indicates that
miscellaneous licencing questions not necessarily related to current
review processes are also considered. Nevertheless I would be pleased
to take this discussion elsewhere if desired.

As one who is committed to free software (in the FSF sense) and open
source (in so far as the two are consistent), I have taken a keen
interest in recent public controversies surrounding the CDDL licence
and licencing more generally. I don't want to bring those
controversies here, except to say that I believe the OSI was justified
in approving the CDDL and that it has the advantages that John Cowan
noted on this mailing list with respect to the MPL, in particular of
being a more reusable licence. Again I would rather not enter into the
CDDL/MPL debate at this point.

First issue: the carefully circumscribed patent grants in the MPPL, which are almost
identical to those of the CDDL (although the latter are slightly wider
with the removal of the "code that you separate" clauses).

Suppose the Initial Developer or a Contributor patents an algorithm,
then implements it in covered code, thereby making a grant under
section 2.1 (b) or 2.2 (b) respectively. You then receive a copy of
the software and (case 1) rewrite the implementation of the patented
algorithm, or (case 2) add a new module containing an implementation
of the patented algorithm. The question is whether you have infringed
the patent contrary to the licence in either or both cases. I can
think of several alternative analyses that might be given of how to
apply the qualifications in sections 2.1 (d) and 2.2 (d) to the grants
in 2.1 (b) and 2.2 (b) in these circumstances, which give different
results, but as I don't have a background in patent law and haven't
researched any relevant case law I'll forego the speculation and just
ask the question.

I also notice that there is an asymmetry between the Initial Developer
and Contributors in both the MPL and the CDDL in that only the latter
are subject to the following provision (here quoting the CDDL version
with the cross-reference omitted):

"You represent that You believe Your Modifications are Your original
creation(s) and/or You have sufficient rights to grant the rights
conveyed by this License."

So, if a contributor contributes modifications in the knowledge that
they are subject to third-party copyright or patent claims which the
contributor isn't entitled to sublicence under section 2, then
presumably the contributor has made an intentional misrepresentation
in virtue of the above-quoted provison. In most jurisidctions would
this amount to a breach of the licence terms, and do you think the
absence of a parallel "representation" requirement covering initial
developers is problematic or are there sound reasons for it?

The CDDL omits MPL sections 3.4 (a) and (b) on the grounds, with which
I agree, that they are burdensome. In an article at groklaw.com the
absence of disclosure requirements in the CDDL is criticized on the
footing that it would allow an Initial Developer to release code under
the licence which is known to be subject to third party claims that
would be infringed by licencees who use or distribute the software.
Contributors couldn't do this without making amisrepresentation under
CDDL section 3.2 (i.e., the provision discussed above and derived from
MPL 3.4 (c)), so the alleged problem only arises with respect to the
original code. As I read it, the Initial Developer also becomes a
Contributor upon contributing modifications to the original software
and is thus subject to 3.2 with respect to those modifications. Surely
the supposed problem could be solved by initial developers disclosing
any third-party interests when the software is first released under
the CDDL. I must admit that I find the objection at Groklaw confusing
to say the least.

An alternative approach to disclosures than that taken by the MPL
would be for open-source projects to maintain a publicly visible Web
page for patent and other disclosures, and to require participants to
disclose their own interests (and/or third-party interests that become
known to them) under certain circumstances. This is exemplified by the
public patent disclosure pages now required to be associated with W3C
specifications under its patent policy. Of course, recipinets of
covered software who weren't contributing to a project that required
disclosure, wouldn't under the CDDL be required to disclose any
third-party claims; but I don't see this as an unreasonable balance.
Obviously this takes us into the area of project best practices and
out of the domain of licences per se.

I like the provisions in the CDDL (section 6.2) that are designed to
discourage the litigation of patent claims. As an aside, the OSL has
relatively straightforward language regarding patent grants that also
covers derived works (the OSL is a very good, well written licence). I
note that unlike the OSL and MPL, the CDDL stands alone in excluding
the contra proferentem principle of interpretation. Finally, I agree
with those who have suggested on this list that it is desirable for an
appropriate organization (whether OSI or some other) to maintain a
short list of recommended licences.

I don't have any background or expertise in intellectual property law
(at law school I specialized in public law - international law, human
rights, Constitutional interpretation etc., after completing the usual
courses in tort, contract, property, Constitutional and administrative
law, criminal law, equity and trusts, which are the core of a law
program in this country).

Regards,

Jason.