Aaron Mulder wrote:
>
> Bear with me - I just want to get this totally straightened out -
> a number of half-completed conversations have gotten me nowhere.
Me too...
> More
> comments below, but the next most important point: how do you feel about
> this section?
>
> "You must cause any work that you distribute or publish, that in whole or
> in part contains or is derived from the Program or any part thereof, to be
> licensed as a whole at no charge to all third parties under the terms of
> this License."
>
> According to that, the GPL doesn't just cover derived works - it
> also covers works that *include* GPL works. A J2EE server package would
> certainly *contain* jBoss, which would mean the whole package (including
> Tomcat or whatever) would need to be GPLd. That would be unpleasant.
Please consider the "mere aggregation" statement that Marc
mentioned. We have no problems if we do "distribute with"
rather than "distribute as a part of". And a jar or zip or
whatever that is used for packaging for the distribution and
has to be unpacked before use can reasonably be considered
part of the medium (as opposed to part of the distributed
contents).
> On Sat, 21 Oct 2000, marc fleury wrote:
> > yes... I disagree and strongly...
> >
> > think a second, it would mean that if we distribute two separate jars on the
> > same page it is ok? (think, does it strike you as stupid?)
>
> Of course it strikes me as stupid. The fact that we have to
> discuss this at all strikes me as stupid, since it is, after all, all
> open-source.
Most legal discussions strike me as stupid and being mostly
a fight about interpretation of laws and agreements and the
meaning of words.
I really would like to blame the designers of our current
legal system, but I don't know how to design a better system.
But discussion on that would be better suited to a list about
politics.
> However, the position I'm being forced to adopt is what is
> defensible in court, not what I think is intelligent.
Makes sense. If decisionmakers think that there might be a
chance the the jBoss license might not hold in court, they
are almost certain to decide against using jBoss.
> > It just says that if parts of the *modified work* are indenpendent and can
> > be considered separate works then you can distribute under whatever license
> > as long as you separate distribution.
> >
> > BUT TOMCAT IS NOT MODIFIED WORK, AND TOMCAT IS NOT DERIVED WORK, it is mere
> > aggregation.
> > So the rest of the paragraph does not apply to Tomcat. It applies to the
> > MODIFIED WORK AS A WHOLE.
>
> Certainly if you run Tomcat and jBoss separately, there is no
> issue about a derived work. But if you run them inVM, there is jBoss code
> that loads and executes "org.apache.tomcat.startup.EmbededTomcat". With
> that, I think the total is a derived work - they are inextricably linked.
Yes, that is derived work.
But this derived work is created at runtime and is not distributed.
Thus the GPL sets no restrictions.
Have to check the Tomcat license to check if it sets any
restrictions on this.
I guess a similar problem exists with all JDBC drivers.
> You can't plausibly argue that "org.apache.tomcat.startup.EmbededTomcat"
> could start a different server. If you created an interface
> ("ServletStartup") which any servlet container could implement, and then
> put the class name in a properties file or something, then I think you
> could say the servlet engine is interchangeable and they are not together
> a derived work. But when the class name is in your source code...
I see. No problem with JDBC drivers, as their names are
in configuration files.
Having the name of the embedded Tomcat startup in a
user-editable configuration file would probably make
a difference to lawyers.
I think that would be nice and clean, and wouldn't
mind even if it isn't required by the Tomcat license.
Best Regards,
Ole Hugaard.
