Hi Eric,
Thanks -- I have been thinking along similar lines. I am also
consulting the author (who teaches at Berkeley, as a matter of fact) of
one of the central legal papers on 'derivative works' and whether or not
dynamic linking creates a 'derivative work'.
Cheers,
Bruce
On 8/18/09 1:56 AM, Eric J Korpela wrote:
> Ahhhh. I get the (significantly more complicated) picture. The
> questions to ask in that case...
>
> 1) Is the copyright individually held or is the copyright held by the
> consortium as an entity? I am assuming the former from your post. If
> so, you would need the permission of any copyright holder whose work
> ends up in the program in order to alter the terms of the GPL. In
> theory you could pare the code down to just what you need in order to
> reduce the number of authors, but that might not solve the problem.
> However, the problem could be worse. I would guess most of the code
> is probably not even owned by the authors. Anything I write at work
> is owned by the UC Regents, although thus far they have been kind
> enough to allow me to set my own licensing terms (within very general
> guidelines.)
>
> 2) Is IMKL or ACML covered by the GPL v2 clause "However, as a
> special exception, the source code distributed need not include
> anything that is normally distributed (in either source or binary
> form) with the major components (compiler, kernel, and so on) of the
> operating system on which the executable runs, unless that component
> itself accompanies the executable." or the GPL v3 clause I posted
> earlier (if the license is "v2 or later")?
>
> If the Intel compiler is required in order to use IMKL and an AMD
> compiler is required to use ACML, you could probably make the claim
> that it is allowed, especially if those libraries are included by
> default in any builds with that compiler. If you can link to these
> libraries using a stock GCC or Visual-C++ compiler, then that claim
> gets more difficult to make.
>
> If the consortium has access to all the authors via email, you might
> be able to get a consensus on an exception. Every 7 days for two
> months send a message: "Some member of the consortium would like to
> add an exception to the license that explicitly allows binaries to be
> distributed when linked with the proprietary optimized math libraries,
> IMKL and ACML. Source for all non-IMKL and ACML modules would still
> be require to follow the original license terms. Please send comments
> or objections to the consortium mailing list within the next N days."
> It might be that universal concurrence is easier to get than we think.
> I would guess a some of the other authors want to be able to use those
> libraries, too.
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
