Re: java bytecode considered bad

[dewar at gnat dot com]

<<I believe it would be trivial, for someone (with commercial interests)
who is able to write a backend, to write a simple dumper that saves the
current state of the internal structures/trees to a file or prints it to
a pipe for consumption of a proprietary backend.
>>

It is not at all clear that this would be valid. If the two halves of
the program had the quality that they only worked with one another, then
given my reading of the precedents, they would probably be considered as
a single program for copyright purposes (The Bentley vs Intergraph case
would have been decisive here if not settled at the last minute, but even
the defense (for which I was expert witness) did not try to claim that
this situation would constitute two separate independent programs -- if
Bentley had won, the situation would be even more protective of the GPL
in this regard.

I don't want to start a long discussoin here, indeed reaching the righyt
conclusion is fairly tricky law, I just want to point out that the facile
assumption that writing a simple dumper of this case would solve the
copyright problems is not necessarily accurate.