OpenJDK Governing Board Minutes: 20011/4/21

[Dr Andrew John Hughes]

On 8 May 2011 21:35, Mark Wielaard <mark at klomp.org> wrote:
> Hi Mike,
>
>> To further set (perhaps lower would be more accurate) expectations
>> on this front, it is my understanding that any revisions to the
>> OCA will happen _after_ the revised Bylaws come out.
>
> That is surprising. The Bylaws are based on the OCA, they use it
> to define OpenJDK contributors and members. They define the inbound
> licensing terms for the project as a whole. Since those are pretty
> essential definitions in the bylaws it would make sense to me to deal
> with them together at the same time.
>
>> Sun, and now Oracle, have business motivations and existing
>> contractual obligations that will always require them to aggregate
>> clear title to the intellectual property in OpenJDK. So anyone
>> harbouring hopes that this would change needs to come to terms
>> with this reality.
>
> The problem is not (business) motivations and obligations themselves.
> All contributors have them. And obviously those are different for
> all of them. There is nobody in the community who is helped by
> unclear titles to any of the "intellectual property" (I assume you
> mean specifically copyrights, distribution terms, patents and
> trademarks) we all work on based on OpenJDK.
>
> To be honest, your response is somewhat offensive. How would you
> feel if some appointed, non-contributor, to the Eclipse board would
> state that anybody harboring hopes that there can be a honest
> discussion on participation agreements, contributor terms,
> inbound/outbound licensing, right grants among members that are
> different from "all rights of all members will get assigned to one
> specific commercial company for unstated business motivations",
> should get to terms with reality?
>
> I think this is a good example of why the current so called governance
> board has a bad makeup. It "governs" based on the motivations of a
> very select group of contributors. Some of which don't even contribute
> themselves. Others are not even bound to the inbound/outbound license
> agreements all other project members uses. These people might be very
> good at defending the motivations and contractual obligations their
> companies have. But there is a very big chance that they completely miss
> the motivations and pain of other members.
>
> As you can expect of the GNU Classpath maintainer, I am actually pretty
> happy about the current legal setup where everybody gets the code under
> the GPL plus Classpath exception, just like with GNU Classpath.
> This makes sure that all contributors, and all end users, get a clear
> and reciprocal license to all copyrights and patents on the code, which
> provides them all the necessary freedoms to use, share, study and modify
> the software any way they like. I think we cannot thank Sun, now Oracle
> enough for that.
>
> But it concerns me that this is not something guaranteed for the project
> as a whole in either the participation agreement (OCA) nor these new
> proposed bylaws.
>
> Not only the uncertainty about the outbound licensing is an issue that
> I think undermines the community aspects of the project. The current
> inbound licensing (only allowed by assigning all rights unconditionally
> to Oracle, without reciprocity) is already currently harming the
> project.
>
> Some of the following issues are not directly the result of the current
> legal setup of the participant agreement and board, but they hurt so
> much more because of the unfairness of the current project setup. And
> because IMHO neither the OCA, nor these proposed Bylaws protect the
> motivations of anybody but Oracle (and their proprietary licensees).
>
> The project could have had a full deployment implementation, applet
> viewer, webstarts, etc. integrated to finish the last few non-free JDK
> requirements. These now live in a separate project (icedtea-web) only
> because the OCA doesn't allow inbound code unless all the rights are
> assigned unconditionally to Oracle (not possible in this case, even if
> the authors wanted, because it is based on some existing free software
> projects). The same was true for earlier efforts of the IcedTea team,
> which were just rewritten by Sun employees because the other efforts
> were based on existing free software.
>
> The project could have actual free (binary) releases/daily builds, since
> thanks to IcedTea we have autobuilders and testers. But when the results
> were offered to be hosted on dl.openjdk.java.net they were rejected
> because of more legal issues.
>
> Worse, the only "releases" OpenJDK makes (like the developer previews)
> are completely unnecessarily under proprietary terms which don't even
> allow contributors to the OpenJDK code to inspect them, nor do they
> allow users to even report issues to the OpenJDK project [*].
>
> There is now support/ports for embedded and alternative architectures
> like arm, powerpc, etc. through the contributed zero and shark code. But
> you still need to use the IcedTea code base, because the in-tree OpenJDK
> versions keep breaking because contributions are only run through some
> proprietary testsuite that don't test the alternative testsuites.
> IcedTea does provide at least some autobuilders for ARM.
>
> Again, worse, it seems Oracle is only interested in some proprietary,
> out of tree OpenJDK port which also happens to support ARM and PowerPC.
> The IcedTea-MIPS port never even got access to the TCK-testsuite.
>
> IcedTea contributors keep trying to push their fixes to OpenJDK in the
> hope to have one real common core free JDK project. But then as recently
> seen on core-libs, their patches are blocked because the Oracle JDK is
> in freeze.
>
> And then there is the constant speculation about the "real" motivations
> of Oracle to harvest all these rights of all contributors. Is it because
> they only really care about their proprietary releases? Will my free
> software project or company be the target of a new lawsuit Oracle
> launches because through the OCA they collect all copyrights and patent
> claims necessary for that?
>
> If you care about a healthy community around OpenJDK then as a board you
> should clarify what Oracle's motivations really are, whether the current
> OCA is the only way to achieve their goals (wouldn't simply registering
> who holds which rights to what code be enough, then they can re-purpose
> code they have all rights to, and simply exclude that over which they
> don't have full rights and/or just accept the same GPL outbound
> licensing all other contributors enjoy) and how to setup things so that
> the motivations of other participants are not harmed.
>
> Thanks,
>
> Mark
>
> [*] More background here:
> http://gnu.wildebeest.org/blog/mjw/2009/11/14/trusting-companies-with-your-code/
>
>

Thanks Mark for such a detailed and clear argument, pretty much all of
which I agree with.  I would clarify one thing, which is when you
refer to ' the only "releases" OpenJDK makes', you are referring to
binary releases.  The source code releases are just zipped up copies
of OpenJDK and are used by the IcedTea project.

I do think the whole situation with the mixed licenses of different
releases by Oracle is something that should have been cleaned up long
ago.  Could someone please explain in what way it is beneficial to
retain this confusion?  Is it something that is being actively worked
on?  Only on Friday, I was speaking to a lecturer at our local
university who thought they had downloaded the OpenJDK source code
with the proprietary builds Oracle provides, and I had to explain that
the lump of code provided there is not the same as OpenJDK.  I could
understand these kind of issues in the early years of OpenJDK, but
four years in, this should have been sorted out by now.

>From the OpenJDK source code base, it looks like the proprietary parts
of Oracle's JDK slot in pretty cleanly to the OpenJDK tree.  If this
is the case, is there any good reason why GPL binaries could not be
provided, with the option to add on the proprietary extensions
(alternative font renderer, colour management, graphics renderer,
plugin, web start) under a separate license?  It seems pretty easy to
me to provide a default which downloads both parts for the majority of
users who don't care, while allowing those who do to just pick the
Free part.

We do a pretty good job of providing a Free alternative for GNU/Linux
distributions, via IcedTea, but users on other OSes (Windows, Solaris,
*BSD, MacOS X) don't seem to have the same level of support.  There is
only http://www.openscg.com/se/ as far as I can see.
-- 
Andrew :-)

Support Free Java!
Contribute to GNU Classpath and IcedTea
http://www.gnu.org/software/classpath
http://icedtea.classpath.org

PGP Key: F5862A37 (https://keys.indymedia.org/)
Fingerprint = EA30 D855 D50F 90CD F54D  0698 0713 C3ED F586 2A37