OpenJDK JRE & Embedded Devices

[hwadechandler-openjdk at yahoo.com]

----- Original Message ----

> From: Mario Torre <neugens at limasoftware.net>
> To: Stephen Colebourne <scolebourne at joda.org>
> Cc: discuss at openjdk.java.net
> Sent: Tue, February 1, 2011 2:05:09 PM
> Subject: Re: OpenJDK JRE & Embedded Devices
> 
> Il giorno mar, 01/02/2011 alle 13.33 +0000, Stephen Colebourne  ha
> scritto:
> > On 1 February 2011 11:30, Andrew Haley <aph at redhat.com> wrote:
> > > All of  us on this list believe that OpenJDK is free software, and you
> > > have  the freedom to:
> > >
> > >    * run the program, for  any purpose (freedom 0).
> > >    * study how the program  works, and change it to make it do what you 
>wish
> > > (freedom  1).
> > >    * redistribute copies so you can help your  neighbor (freedom 2).
> > >    * distribute copies of your  modified versions to others (freedom 3).
> > 
> > "All of us on this  list believe that OpenJDK is free software". No,
> > I'm afraid I  unconvinced that Oracle's lawyers/managers necessarily
> > agree, at least  in the way that the point was phrased above:
> > - Considerable efforts have  been undertaken to prevent an open source
> > implementation of Java SE from  running on embedded devices
> > (Harmony/Android)
> 
> Java, yes. We're  talking about OpenJDK here, though. I know at least two
> cases where OpenJDK  has been put and sold on different embedded devices.
> 
> So, this is possible  as soon as you *comply* with the *license*.
> 
> The corner case is hotspot,  but if I remember correctly Cacao managed to
> get the TCK running a mixture of  Classpath and  OpenJDK:
> 
> http://openjdk.java.net/groups/conformance/JckAccess/jck-access.html
> 
> So  there is at least at least an historical precedent.
> 
> > - There is an  active legal case (vs Google).
> 
> I don't comment on this here, but I  believe that, unfair as it is, if
> Google used OpenJDK they would have avoided  this. We have no proof of
> this though, so it doesn't matter.
> 
> > -  There are active and known patents claimed (see Google case).
> > - Oracle  doesn't necessarily respect legal & social agreements,
> > notably when  the other party is a not-for-profit (Harmony/JSPA)
> 
> Well, this is a bit  exaggerating, they are not super legal entities...
> 
> > - The strength of  the GPLv2 has yet to be fully tested in court IMO.
> 
> Nobody in the world  ever violated the GPL so that no legal action was
> ever required? Wow, so  probably the GPL alone is indeed enough :)
> 
> > Some consider the above  to be FUD, I consider it to be healthy
> > scepticism based on how the  Oracle is running the ecosystem. YMMV.
> 
> It could be legitimate fear, which  I partially share, but the way you
> put it is really just FUD,  really.
> 

The OpenJDK license and agreements clearly state that one can create any  
product as a derivative of OpenJDK. If one wants to get a JCK license  and make 
claims of Java compatibility then they need to not modify  things considerably 
and there are certain things which can't be changed or touched. Otherwise, and 
outside of compatibility, one can do  much of what they want with the openjdk as 
long as the derivative is  licensed according to the terms of the openjdk 
license. That means one  can even distribute and link to that derivative as 
needed per the  openjdk license.

The keys are 1) being a derivative 2) do you want to claim "Java" compatibility 
or not 3) do you want to use the trade mark OpenJDK outside of "fair use" only 
terms and if so must not change the derivative beyond a majority. 2 and 3 are 
really optional based on use etc.

IANAL but some of this looks pretty obvious.

Apache Harmony and Android are two completely different issues from OpenJDK 
derivatives (with changes or not), so those things really are separate. There 
are clear words in the licenses of the Java Specifications, but then their 
origin, the JCP and the agreement the JSPA should invalidate those encumbrances, 
but of course on the Android issue if Google copied and pasted some code...their 
bad:

Just a bit of a blabber and rant here. Please ignore if you don't want to read a 
long email:

1) The issue with Harmony isn't whether they can distribute the software; well, 
it is and it isn't per the use of the TCK and field of use and the JCP and JSPA 
along with prior public statements made by Sun; would be a legal battle. 


This is because it is a clean room implementation apparently, and with OpenJDK 
they explicitly say one can create derivative works (including those which 
change things considerable as long as they don't claim compatibility with "Java" 
nor run or get the TCK for it). I think what Sun and now Oracle is doing in that 
regard, considering their own processes, documents, and statements prior to 
Harmony, is completely wrong. Were a valid court to exist which understood all 
this, invalid meaning courts don't have the expertise in house without external 
counsel to try to make the best decision they can on such subject matter, and 
someone willing to test how well this would hold up in court, then I think that 
would be cleared up, but given the parameters at hand, it seems nothing is going 
to happen there. 


My personal feeling is Apache should just release Harmony without that 
certification and not make the claim about Java and compatibility...not even 
call it Java perhaps, but I think Sun/Oracle's own claims and statements would 
be used against them were they to try to take Apache to court for any use of the 
word Java or compatibility even without running the TCK (their "planned" license 
terms...i.e. what folks were voting on at the time or probably more correctly 
the information which led to groups working on the JSR to its completion etc; 
was work on the spec obtained by unjust means if prior to license release 
statements of good faith were publicly made or in a manner which can be 
verified). Let's say Harmony was 6 versus Java 5:

From: http://jcp.org/en/jsr/detail?id=270:
"7. Nothing in the licensing terms will prevent open source projects  from 
creating and distributing their own compatible open source  implementations of 
Java SE 6, using standard open source licenses. (Yes,  you can create your own 
open source implementation of Java SE 6 if you  really want to. But we're also 
doing everything we can to make it easy  for you to use the original RI sources! 
See http://jdk6.dev.java.net.) "

"10. The Java SE 6 TCK will be offered for license at no charge,  with trademark 
and branding rights, but without support, to qualified  not-for-profit entities 
(including not-for-profit academic institutions)  and qualified individuals 
engaged in efforts to create compatible  implementations of the Java SE 6 
Specification."

Understandably Apache is in a tight spot though... They have other projects etc, 
and law suits cost money. That money affects the other operations; too much and 
it could kill Apache. The above statements, Oracles prior statements in 
Harmony's defense, and the JSPA seems like a hell of a case.

Too, I think if one looks how Apache is structured there is money coming from 
different companies which probably influence some of that too; different 
employees involved, different CEOs pulling wallet strings, etc etc. I can't say 
for sure on some specifics, but there are certainly different interests 
involved. Different Apache devs and members signing NDAs etc and working for 
competitors. Can cloud some things even more; such as who is really doing X for 
the greater good versus a competitive edge and who can trust who and how common 
ground is achieved. Just being real here.

I believe there are some clear violations on Sun and now Oracles part which 
would play out in Apache's favor in courts. Good faith statements matter, and we 
have a lot of that taking place in different capacities from both Sun and Oracle 
in the past. Of course, this all depends on which state one were to sue as well, 
and this would have to be California, USA.

2) Android Dalvik is not a JDK nor JRE. It is a different thing altogether which 
converts binaries. I think at the end of the day, based on Oracles own 
statements before purchasing Sun in relation to Apache Harmony and other things, 
some of what occurs with OpenJDK, other things related to be able to implement 
Java specifications as long as you don't mention Java or run the TCK etc the 
case will be very hard to prove one way or another with regard to patents if 
those statements are allowed to be entered as evidence along with some of the 
above from the harmony section.

Now, what could get them in the hornets nest so to speak is if there is a line 
by line copy of some code some where. Obviously this would have to be technical 
code, but too, courts don't have the skills required to really sort this 
out...legal folks are smart in their areas, but when it comes to CS, if they 
don't have that background, they can't really get it, and that could make 
technical code versus "public String toString()" or something harder for them to 
differentiate.

That said, that part of the issue is manageable with regard to exposure if it 
can be proven that steps taken by Google affected Oracles wallet directly; i.e. 
were damages done. Damages etc would be paid along with costs and penalty, code 
would be changed, probably a cease and desist would be in place until the court 
is satisfied. Hopefully Google didn't do that. If they did, then it will be more 
cloudy; carriers will probably drop them etc and that could likely kill Android 
or cause a slow death of developers fearing no indemnification if they continue 
outside a cease and desist. 


Too, does the Dalvik have anything from the JVM in there. Are we just talking 
about the converter/compiler versus the runtime. That will all just have to play 
out.

But, something I can see here is perhaps a shifting of the winds were Google to 
win. It would be a game changer for Apache and all I believe. Precedence would 
be set. New parameters would be visible for all. I think the JCP and some other 
things would get cleared up if their prior statements came to be the Achilles 
Heel.

All this just really points to issues with software patents; leaving out whether 
or not copyright issues were violated or not; definitely copy and paste is a bad 
thing. I mean, what parts of Java were inspired by reading specifications for 
other languages? What keywords came from some other specification? What 
libraries ideas were copied from some other system? On and on. It is like 
patenting nuclear processes. We don't create those things; we just learn how to 
understand them. Same with applications or that damn one click bid thing someone 
patented ;-). If we continue on such a course I can envision a time when writing 
a program would be more about what one can legally implement versus getting the 
work done. Yeah...that economy will work out for us.

Wade