[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: GPL, yet again. (The kernel is a lot like a shared library)

** Andrew Suffield ::
> On Thu, Sep 08, 2005 at 01:22:07PM -0300, Humberto Massa Guimar?es
> wrote:
> > 3.3. it seems to me that it's absurd to think, for instance,
> > that Debian cannot dynamic link a GPLd program with OpenSSL.
> > Why? Because if I write a completely-compatible MassaSSL library
> > and install it in my system just in the same
> > places/names/sonames/whatever OpenSSL is installed, this would
> > change the copyright status of _the_ _program_!!
> This says that there can be no such thing as copyright
> infringement for creating a derivative of a piece of software,
> because you can always replace the original with a
> reimplementation that wouldn't be infringement.

my knowledge of the English language is still worse than I tought,
because I do not have any recollection of meaning what you said _at_


> While it may be an interesting legal theory that copyright
> infringement in software can only apply to verbatim copying (and
> one that has been proposed before by various crackpots), I would
> not like to rely on it in court, because it's absurd.

I have not said _at_ _all_ something absurd as you state in your
paragraph above. Calling me (directly or indirectly) a crackpot will
not change this. I do know what I'm talking about and I think you
are not being polite.

> I'll leave it as an exercise for the students to find where the
> argument went wrong; the mere existence of a flawed conclusion is
> enough to convince me that it went wrong *somewhere*. Go back and
> do it again in a manner that concludes derivative works are
> normally infringement, and explains why this case is different.

The problem here is that you got to a flawed conclusion that I did
not state at all, so the error resides entirely in your parser,

You seem to be saying that if a program has

  x = MD5(a, b, c, d);

in its text, then it *is* a derivative work of OpenSSL. What I said
is exactly the opposite of this: the presence of the above line does
not imply at all that some program is an OpenSSL derivative. And
it's not. The usage of a library by a program does NOT transform
said library in ANY aspect. My point is exactly that, throughout all
my argument: "a work based on the Program" <==> "derivative work
under copyright law" <==> "the result of an intellectualy-novel
transformation applied over the original work".

"Usage of a library, especially thru its published API" !=
"transformation applied over the library"
"usage of a library" != "a work based on the Program"
"usage of a library" != "said library must be GPL-compatible"

Maybe my English was clearer now?

Mind you, the problem with ike-scan (today on-list) was exactly
this: it (optionally) calls MD5 and/or SHA1 from OpenSSL, offering
an alternative implementation if OpenSSL is not available
(OpenSSL's implementation being better performant). No court that I
know of would regard ike-scan as being a derivative work (nor a
"work based on") OpenSSL. And I had my share of court work.


Reply to: