harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dalibor Topic <robi...@kaffe.org>
Subject Re: NDA issues and acceptable use of sun source
Date Mon, 13 Feb 2006 13:26:43 GMT
Leo Simons wrote:

>>In absence of court decisions, there is just the possibility to draw
>>very clear lines what constitutes safe contributions and what doesn't.
> I disagree that this is possible. Combining intellectual property laws
> from a variety of jurisdictions with many years of open source and closed
> source history means that there is no "safe" and there is no "very clear".

Not looking is a very clear, bright line. You can't infringe copyrights
on works you don't access.

When we move away from that, we have to evaluate probabilities of
infringement, and enter the world of "* enough". Whether a case is "*
enough" largely depends on the case, what access has been made, what
legal arrangements covered it, and what part of the contribution may be
covered by those legal arrangements. We can draw some rather clear "safe
enough" lines, where we can reasonably hope to persuade judges, if it
becomes necessary.

For example, if a contributor only ever looked at Sun's pre 1.0 code,
and wanted to contribute an implementation of java.util.concurrent, it'd
be hard for anyone suing us to argue that we'd infringe their copyright
on something the contributor could not have possibly accessed in pre 1.0
code, since it wasn't there in the first place. No access, no infringement.

Of course, the other party could argue that the contributor breached
some contract with them, but that's between the contributor and whoever
he has contracts with. Given the plethora of Sun's licenses for Java
technology in the past 10 years ... way too much work for anyone but the
contributor to figure out, since the actual license texts change all the
time subtly (JRL is now at 1.6, for example).

> Anyhow. I feel that Harmony should not have a policy as strict as
> Classpath ("if you ever looked at sun source, you can't contribute"). I

The major difference between the two is that Classpath does not want to
have to deal with the probabilities. Mandating that people don't look is
a pretty good way to do that, as explained above.

Otoh, Harmony needs to weigh the probabilities, if it aspires to include
runtime developers who've been exposed to sun's source, so that means
making educated guesses.

I think Geir's policy document is a pretty good one for that goal.

> think that it is absurd if guys like Tor can't contribute a vorbis
> implementation (vorbis being something explicitly designed to be very
> free of legal mess, mind you) to an open source project just because 10
> years ago they looked at source code that had nothing to do with vorbis
> (which didn't exist at the time in any form!). 

The underlying issue is pretty simple: was there something he could
inadvertingly copy in the proprietary code bases he studied into his

If no, great, we're game according to Harmony CLA rules.

If yes, it's a tough call, and needs to be checked, for example by
examining what the contributor studied, and whether those bits he
studied are similar to his contribution.

Sure, dealing with proprietary software is frustrating. People who've
entered those contracts back then surely felt that they were worthwhile
with all their consequences, though, and made those choices voluntarily.

Unfortunately, we can't help them retroactively change the consequences
of their choices: figuring out the precise legal status of their
contracts/NDAs/obligations is up to contributors, and whoever they have
contracts with to figure out.

dalibor topic

View raw message