Re: mini vote

[Tom Tromey]

>>>>> "Mark" == Mark Wielaard <address@hidden> writes:

>> On Wed, Jul 18, 2001 at 10:45:31AM -0400, Stuart Ballard wrote:
>> Transvirtual has been good to the free software community, so we don't
>> want to interfere with their business model". Personally, I don't think
>> that just because a company has been good to free software entitles them
>> to a perpetual "non-compete" agreement from the rest of the community -

Mark> I do agree with you here, but why should we now switch
Mark> companies?  Cygnus/RedHat has been good to the Free Software
Mark> community but they also didn't license their implementation of
Mark> the AWT under the LGPL.

Let's be more clear about this.

I agree that the FSF should not make decisions to support any
company's business model.  However, that isn't what happened with
libgcj.  My understand and recollection is that two things happened:

* Cygnus wrote libgcj from scratch and donated it to the FSF on the
  condition that it would be under the GPL+exception.  Yes, this was
  to support Cygnus' business model of selling gcj+libgcj to embedded
  developers.  I don't really see a problem here.  Cygnus didn't have
  to donate this code (in fact it took us developers a long time to
  convince management to do it).  The FSF didn't have to accept it.

* gcc has a policy of requiring runtime libraries to be usable by
  anybody.  For instance, the C++ standard libraries are under a
  license suitable for use by embedded developers.  Presumably the FSF
  supports this policy (I'm not involved with policy setting for gcc)

Saying that licensing libgcj in this way supports only Red Hat denies
the reality that anybody is free to build their business doing gcj
support.  People have already done this with g++ and gcc.
Transvirtual uses gcj -- so in effect, work we do on the compiler
supports the very company that we can't "undercut" (RMS' word) by
relicensing AWT.  This is ironic, don't you think?

Mark> Since the AWT uses native peers that are implemented using GTK+
Mark> which is already licensed under the LGPL I don't think it
Mark> matters that much if the java portion is licensed under the LGPL
Mark> or GPL+Exception.

It matters if you're interested in writing your own peers which, say,
write directly on the frame buffer.

Mark> But if that license will be the GPL+Exception we better make
Mark> sure that everybody (including people that doubt the language
Mark> that is used for the exception currently) is happy about it.

I don't really understand the legalities of the exception.  It was
chosen because libgcc used it for a long time.  I'm not sure how we'd
go about changing it, but I personally am open to the possibility.


Should the Classpath AWT be relicensed?  Obviously we'd like to have a
working AWT in libgcj.  But on the other hand it isn't worth it, to
me, if it causes a lot of contention and bad feelings about the gcj
project.  We'll eventually finish our AWT implementation.  It isn't
entirely clear that changing the Classpath license would make this
happen more quickly.

Tom