--vkogqOf2sHV7VnPd
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Thu, May 05, 2005 at 09:10:41PM -0400, Tim Kelly wrote:
> On 05 May 2005 20:27:55 -0400
> "Nathan J. Williams" <nathanw@wasabisystems.com> wrote:
>=20
> > If using ideas from A in B always causes B to be a derivative work of
> > A, then nobody who has knowledge of GPL'd code can, from a paranoid
> > perspective, ever produce code that they can prove is not a
> > GPL-derivative work.
>=20
> "Knowledge of" and "knowledge derived from" are not the same.  I have

You may be correct about the two forms of knowledge being different,=20
however that does not change the fact that neither of them is=20
copyrightable.

To quote http://www.copyright.gov/circs/circ1.html#wnp (note the .gov,=20
this is the US copyright office; they should know what they are talking=20
about):

Several categories of material are generally not eligible for federal=20
copyright protection. These include among others:

[item about works that have not been fixed in tangible form and item about=
=20
short words and names (i.e. things that get Trademarks, AFAICT)]

* Ideas, procedures, methods, systems, processes, concepts, principles,=20
discoveries, or devices, as distinguished from a description, explanation,=
=20
or illustration

Oh, even better. US Copyright law circular 92 Chapter 1 section 102(b):

(b) In no case does copyright protection for an original work of=20
authorship extend to any idea, procedure, process, system, method of=20
operation, concept, principle, or discovery, regardless of the form in=20
which it is described, explained, illustrated, or embodied in such work.


Thus as long as you only gain ideas & such, as distinguished from the=20
specific work they are produced in, then you have gained something which=20
can not be copyrighted and thus you are not infringing on a copyright.

> knowledge of how Paul Mackerras wrote the mace driver for Linux on Apple
> hardware. I haven't learned anything from it, unless Dave Huang based
> his driver code on it, and from the archives it appears he did not, nor
> are they similiar.  There is enough documentation available from AMD to
> learn what Dave did, and I've gone through it and verified everything he
> did (nice work, by the way, I learned a lot). There is also non-licensed
> documentation on the CHRP standard that explains dbdma in detail.=20
> Therefore, the mc* macppc driver is not derived from GPL code.
>=20
> The courts have regularly acknowledged literary material may have
> similarities and overlap extensively without one being derived or
> copied from the other.  This is because the products were created
> independently while utilizing the same common knowledge.  Your
> example, that of the ADB keyboard, is the exact opposite.  There is one
> and only one source of knowledge about that keyboard.  You can not
> derive that knowledge independently of Apple's code.

As above, the knowledge is not covered by a copyright. It may be covered=20
by non-disclosure agreements or patents, but not copyrights. Thus while=20
other factors of law may come in, copyright law doesn't apply.

> Your interpretation of "derived" means that if I "or" several bits on
> separate lines instead of on one line as in the code I have viewed, it
> is not "derived," even if I have no way of knowing why I should do this
> other than I saw it in the original code.

Please do not put words in Nathan's mouth.

If you are cerating your own work expressing the knowledge you gained, as
opposed to copying the other work, then yes, it is not an infringement of
the copyright on the original work. If however it looks like you merely
took the other work and made a minor edit of it, then that'd be an
infringement. A rather clear one at that. :-)

I think this came up recently. The way I think about this is consider what
you would do as a student or grader in a computer science class. How do
you know if a homework is the result of cheating? If the project is to
write a B-tree routine set for a specific implementation type (say
red-black), well, all the homeworks are going to look alike; there's only=
=20
one method to use. The graders in the classes I took were able to detect=20
cheating, as even though the method had to remain the same, there was=20
enough individual expression in the work that copying (cheating) was=20
detectable. Thus it is possible to have multiple original works that=20
embody the same methods, ideas, procedures, porcesses, concepts, and=20
princimples. ;-)

> > It has to be a secret to be a trade secret. Publishing the source
> > under a copyright license is entirely not sufficent for maintaining a
> > trade secret.
>=20
> Not according to the DMCA.  All it has to be is something that the
> company uses while doing commerce that would cause a loss of
> competitiveness if another company obtained that information.  The
> information from Apple was not released as public domain.  If NetBSD
> uses knowledge gained from APSL code that then causes Apple to lose
> sales of OS X, and NetBSD is not complying with the APSL, Apple can sue
> under the DMCA, copyright law not-withstanding (der Mouse ought to
> appreciate that last clause).  The DMCA does trump copyright law; that
> is one of the reasons why it was written.

Can you please quote sections? I have reviewed the Copyright Office=20
Summary of the DMCA, and I can find no such coverage. I also downloaded=20
the DMCA, and further found no such provisions. Note also that the DMCA=20
does not trump copyright law, it amended existing copyright law, and now=20
is part of the current US law.

There is a discussion of sections regarding circumventing the steps
copyright owners have taken to protect their works, but this is aimed at
preventing breaking digital rights management systems. Apple has released
APSL code under no digital rights management systems (it's not like you
have to have iTunes or Xcode decode the encrypted source files). So I do
not see how that applies.

> > > You're advocating a position that you are not offering legal
> > > means to defend.  Ask Wasabi Systems' legal department if they

As are you.

> > > are comfortable with your position and how you obtain information
> > > that you then incorporate into a product they offer in competition
> > > with Red Hat, and also whether they are willing to go to court with
> > > your position.
> >=20
> > Been there, done that. Wasabi Systems looks at Linux (and other OSS)
> > code all the time.
>=20
> Make sure your clients know this.

Do you really think Wasabi would permit looking at Linux if our legal=20
councel felt it would put the company or our customers at risk? I know our=
=20
lawyer, and he was one of the founders, so I don't think he'll get this=20
wrong and I don't think he'd risk the company.

Tim, you're spreading a lot of FUD about copyrights. Do you have a legal
brief outlining the position you're describing? Because folks who are FAR
more paranoid about this than either you or I are (and have been at this
longer than I expect either you or I have known about NetBSD) have looked
at this question in detail. And the resulting opinion is in line with what
Nathan and I and others have been saying; gaining knowledge from another
piece of code does not infringe on the copyright of the original work.=20
Copying it does, but not using the knowledge you gain.

Take care,

Bill

--vkogqOf2sHV7VnPd
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (NetBSD)

iD8DBQFCe+zfWz+3JHUci9cRAvnkAJkBd0/MQiTRmDEwU7DG8Tw4VXh1hgCdEdr5
jboztrPW8sElwO6Mr4H5b6Y=
=CG9t
-----END PGP SIGNATURE-----

--vkogqOf2sHV7VnPd--