If the only part "borrowed" is a non-expressive,
and hence
apparently (in the EU) non-copyrightable aspect,
then it would not be a
derivate work, so the copyright owner
would have no leverage over the client
program. If the
copyright owner has already licensed the GPL code to the end
user, where is the violation?
Yep, I think I was overreaching
there. Unless the license
of the client program essentially removes some rights
that
would have been granted for the library, I don't think
there'd
necessarily be a violation at that end.
If using the library or its
headers as a code development
tool places constraints for the resulting
deliverable and
those constraints are not met then that's where the
violation
is.
Whether such constraints have any legal weight is another
question,
but bypassing them does indicate that the license
is not agreed upon, so any
rights to the library would
be those given by copyright law. It is in this
context that
the "stating of facts" starts to seem odd, as in "if you
want
this program to work, you need a library providing
these functions present
under this filename - I know of one
but
won't provide it because I don't agree
with the license."
Still, I don't see the GPL sticking to the API, just
the
implementation behind it. Wasn't that the whole idea behind
the
BSD-licensed BusyBox clone floated earlier?
JK Finn [ Reply to This | Parent | # ]
|