Re: Some missing facts (Was: Eclipse 3.0 Running ILLEGALY on Kaffe)
Etienne Gagnon wrote:
Regarding the Kaffe FAQ at:
http://web.archive.org/web/20011211201322/http://www.kaffe.org/FAQ.html
In this document, it is clearly written:
Is the information given in this FAQ binding?
The information in this FAQ is accurate to the best of our knowledge,
but there is no guarantee. It is drawn from public and private
statements by Tim Wilkinson and others. If you need binding information
about copyright issues, contact the copyright holders.
So, in other words, you CANNOT interpret the text of this FAQ as an
exception to the GPL. This FAQ is in now way legally binding, and it
has not even been written by its copyright holder(s) [or, more
precisely, agreed upon by all of them].
None of the the things you are trying to assert are legally binding
either, as you are no copyright holder of Kaffe. No copyright holder of
Kaffe has ever attacked Debian or another distribution with bogus
copyright claims.
What's your point again?
Fact 2
======
In its early years, the Kaffe license was changed from BSD-like to the
GNU GPL, explicitly to force people to get into separate licensing with
Transvirtual, when the terms of the GPL seemed too restrictive.
The FSF did not want its own project, GNU Classpath, to harm the
business of Transvirtual which was making an important contribution to
Free software by releasing Kaffe under the GNU GPL. So, for many years,
GNU Classpath's java.awt.* package was actually licensed under the GNU
GPL *without* the linking exception. The reason was, explicitly,
because the FSF believed that doing otherwise would undercut a stream of
revenue for Transvirtual. I have learned of this through private email
discussion with Richard Stallman, but there exists also some very public
evidence of this:
http://www.mail-archive.com/classpath@gnu.org/msg02052.html
http://www.mail-archive.com/classpath@gnu.org/msg02050.html
That's not because of the interpreter.
Transvirtual had a specific, hacked up version of gcj for Kaffe that
allowed them to precompile the class library and applications, afaik,
for embedded systems. In that case, as I've explained before, the
copyrightable parts of GPLd class libraries would end up in the compiled
works, and require compiled works to be licensed under the GPL. It ran
pretty fast and allowed Transvirtual's Kaffe to be used on mobile
devices, afaik, like Helios. Presumably that was supposed to allow for a
Trolltech like business-model. Well, it failed ;)
Work on this feature had been done for one specific version in Kaffe's
CVS back in 1999, according to http://www.kaffe.org/doc/kaffe/FAQ.gcj.
It predates your links, btw.
Do you have any proof of a copyright holder on Kaffe
enforcing this bogus interpretation of the GPL of yours on anyone
or is this all based on hearsay?
I think that the first paragraph is quite explicit.
So is the exception.
Kaffe's class library, as a whole, must be licensed under the terms of
the GNU GPL or not at all. The GNU Classpath linking exception cannot
be applied to the class library as a whole, unless Kaffe's copyright
holders (all of them) agreed to add this exception to the license of the
classes under their copyright.
Irrelevant to this discussion, as a program using standard Java
interfaces is not automatically a derived work from Kaffe, not matter
how you try to twist it.
Andrew Suffield quite effectively debunked your 'Kaffe's GPld portions
of class library library automatically infect everything Java
application they touch' nonsense in November 2003 already right here on
debian-legal.
http://www.mail-archive.com/debian-java@lists.debian.org/msg03572.html
http://www.mail-archive.com/debian-java@lists.debian.org/msg03575.html
Are we done now?
cheers,
dalibor topic
Reply to: