GPL with the Classpath exception - clarification needed

[John Cowan]

Wilson, Andrew scripsit:

> Not quite with you here. Agreed subclassing does not *modify* the base
> class.  However, are there possible circumstances where a subclass is
> a derivative in the copyright sense?  If the answer is yes, the rest
> of your analysis is moot.

I don't see how it could be a derivative work, any more than a function
that calls your function is a derivative work.  To be sure, the resulting
executable as a whole is a derivative work, but the subclass itself is
not, and so out of scope for the LGPL.  Analogously, if you put forth
an argument and I answer you point for point, my argument is in no way
a derivative work of yours, though it might be hard to follow without
your work.  (If I incorporated the relevant parts of your argument,
as in this very email, the combination would be a derivative work.)

> Such a subclass, or instantiation, is not a "work that uses the
> Library."  It is a "work based on the Library," and it is LGPL.
> The only exception for a derivative under copyright is the familiar
> #include rule.
> 
> As LGPLv2.1 helpfully notes, "the object code for the work may be a
> derivative work of the Library even though the source code is not.... "

Quite so.  Under the GPL, a derivative executable work must be distributed
with full source.  Under the LGPL, you only need to distribute the
LGPLed source.

-- 
John Cowan  cowan at ccil.org  http://ccil.org/~cowan
If I have seen farther than others, it is because I was standing on
the shoulders of giants.
        --Isaac Newton