Re: testing before a release

[Etienne M. Gagnon]

On Wed, Nov 07, 2001 at 12:06:51PM -0800, Per Bothner wrote:
> 
> Note the GPL + exception is only a permission to "reuse contributed code 
> without
> any source-code redistribution obkigation" if you "link this library". 
> If you take
> contributed code and use it apart from the library, or if you modify the 
> code in the
> library, then the GPL applies.

Ah hah!  This is quite different than RMS's interpretation.  He wote to me (in 
a 
message CC'ed to both Brian and Paul) that if I write derivative work, I have
the choice carrying the exception text into the modified version.

If this is not possible, it means the following:

If I modify Classpath code to adapt Classpath for my VM, then Classpath
becomes GPLed, and thus my VM has to be GPLed as all applications running on it,
(e.g. I cannot run the SPEC benchmarks, javac, or any non GPL compatible app).

If this is really the case, I will stop using Classpath.  I think that Intel 
will
also be very interested to know about this restriction, as they probably 
modified
Classpath a little so that it works with ORP.

If this is really Classpath's objective (only FSF copyrighted code can benefit
from the exception, no other bug fixes unless contributed to and accepted by
the FSF), then Classpath might not be usable by many Free VM developers.

Now, you will agree that if the exception can be carried over to derivative
works (as RMS seemd to think), then it is a huge hole.

So, no matter how you interpret the GPL exception, there IS a problem.
This is exactly why I am suggesting to change the license to LGPL + exception
that eliminates the dynamic link (or .obj) requirement.

So, do you think my proposal makes sense?

Etienne

-- 
Etienne M. Gagnon                    http://www.info.uqam.ca/~egagnon/
SableVM:                                       http://www.sablevm.org/
SableCC:                                       http://www.sablecc.org/