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.

Reply via email to