[License-discuss] Views on React licensing?

[Henrik Ingo]

On Fri, Dec 2, 2016 at 6:26 AM, Richard Fontana <fontana at opensource.org> wrote:
> - is it good practice, and does it affect the open source status of
>   software, to supplement OSI-approved licenses with separate patent
>   license grants or nonasserts? (This has been done by some other
>   companies without significant controversy.)

This should of course be discouraged. However, I sympathize with this
kind of setup if it is intended to be a proposal for a license that
doesn't yet exist. If Facebook a) intends for the combined license to
qualify as open source, and b) eventually submit it for OSI approval,
then it seems to me this is a natural path towards such a goal.

> - does the breadth of the React patent termination criteria raise
>   OSD-conformance issues or otherwise indicate that React should not
>   be considered open source?

My view is clearly no. To me these objections seem similar to people
claiming that GPL is unfair, not not as free as permissive licenses,
because it does not allow them to use GPL licensed code in certain
ways that they would want to. Tough luck, you don't get my sympathies.

Technically it seems to me the argument against the React patent grant
is that a license must only list conditions applying to the specific
piece of software at issue, e.g. contained within a single github repo
or tar file. I note that in the reality of today, this opinion seems
to purposefully argue for as narrow a patent retaliation clause as
possible. Possibly the people (which I don't know) arguing this has
such an agenda?

The reality of today is that software is commonly distributed in very
small pieces. A framework like React is likely to grow not as a single
repository, but as an ecosystem of multiple small modules (was it
something like trim() that caused such a mess in npm recently?), each
version controlled and distributed separately, each with their own
license file, however, most of them using the same license. It is IMO
a perfectly valid view to say that when you participate in this
commons as a whole, such as by downloading and using for free multiple
modules of such a commons, you must enter into a patent truce with
that community as a whole, not just refrain from suing the particular
implementation of some trivial function like trim().

We should also consider that the narrow view would in fact favor more
closed source companies suing companies publishing lots of software as
open source. (E.g. Oracle vs Google, or Microsoft vs Everyone).
Companies not publishing (majority of their software as) open source
would enter patent disputes with a stronger hand, since they don't
give away anything themselves, but their patent aggression targets
have given away right to counter sue for all of their IPR except at
most for some specific piece of code at issue. IMO it is in the
interest of OSI and the FOSS community to allow terms where those
companies actually contributing open source software have an equal or
even stronger position in patent suits.

In addition to the fact that OSI has historically approved such patent
clauses, I'd also want to list as "similar in spirit" precedent the
AGPL §13. To achieve its activist goal, the AGPL requires you to
distribute not just any AGPL code covered by the license, but also
regular GPL code if used in the same work.

Speaking of GPLv3, at the time it was drafted I remember I was hoping
FSF would put into it similar or even stronger "global patent truce"
kind of patent retaliation clause and I was disappointed it didn't. If
you ask me, it would be within the OSD for GPL to say essentially: If
you assert patents against any GPL licensed software, you've lost your
license (maybe even copyright license?) to use any GPLd software from
anyone in the world. Probably my position is rather extreme then?

> - if the React patent license should be seen as not legitimate from an
>   OSI/OSD perspective, what about the substantial number of
>   past-approved (if now mostly obsolete) licenses that incorporated
>   patent license grants with comparably broad termination criteria?

Such a view would therefore be contradictory with precedent.

> - should Facebook be encouraged to seek OSI approval for the React
>   license including the patent license grant?

Yes. As far as I can see, the BSD + additional stuff should be a
single file and single license, and OSI approved.

henrik
-- 
henrik.ingo at avoinelama.fi
+358-40-5697354        skype: henrik.ingo            irc: hingo
www.openlife.cc

My LinkedIn profile: http://fi.linkedin.com/pub/henrik-ingo/3/232/8a7