On 5/28/07, Emil Volcheck <[EMAIL PROTECTED]> wrote:
> I've been spending my vacation this weekend at Balticon,
> the Baltimore Science Fiction convention (See http://www.Balticon.org/ ).
> I spend most of my time in the science track, and one of the
> speakers was noted open source guru, Eric Raymond, author
> of "The Cathedral and the Bazaar". There was lively discussion.
> I asked him what advice he had for me regarding the choice of
> license for open source computer algebra systems, and what
> SIGSAM should promote: GPL or BSD/MIT? He came down firmly
> in favor of the BSD/MIT model.
>
> I wasn't too surprised, because I know that he and Richard Stallman
> are kind of polar opposites in the open source software community.
> I don't necessarily buy Eric's argument, but it's interesting. In a
> nutshell, he said that the benefit of open source developers allowing
> corporations free use of software under a BSD/MIT license brings
> greater benefit to the community and outweighs the fear that some
> companies have of the GPL (e.g., its "virality"). Eric believes that open
> source is inherently a better and more economically efficient mode of
> software development than traditional closed source development.
> Eric believes that companies would be wise to adopt open source
> development, but that each company has to decide whether there
> is a business case to be made in each case.
>
> SIGSAM needs to think about what role we should play in this arena.
>
> Perhaps I should add that I'm a member of the Free Software
> Foundation. But while I think the GPL is a brilliant invention, I can
> see there are times when there is more value created by using the
> Lesser GPL (or perhaps even BSD/MIT).
>
SUMMARY of my answer:
GPL versus BSD for open source computer algebra software
is a choice that was already made by the previous generation
of mathematicians. It's too late now to change things.
More details, for those who like such things:
It must have been exciting to talk to ESR.
Unfortunately, he is wrong about which license to choose, at least
in regards to SAGE. The GPL is the correct license choice for
SAGE, because the key mathematical software SAGE is built on is
Maxima, PARI, Singular, GAP, NTL,
and these are all GPL'd (not LGPL). Because of the viral nature of the GPL,
if the SAGE developers wanted to go the BSD/MIT route, we would
not be allowed to build on Maxima/PARI/Singular/GAP, and
would have to reimplement the functionality of Maxima/PARI/Singular/GAP/NTL.
Reimplementing the functionality of these systems
is not an option because it would set us back 10-20 years, and -- more
importantly -- even if successful it would fracture the open source
math software community instead of unifying it.
I do understand that new SAGE code could be licensed BSD
and use Maxima/PARI, etc., but that's pointless since the GPL
would still apply to SAGE and any derived works. I've already
experienced this -- the Scipy/Numpy community avoids SAGE and
the code we've developed for SAGE (e.g., the notebook and
interfaces) because it's GPL'd and they work super-hard to
make sure 100% of their codebase is compatible with the
BSD license (which gives the Enthought company lots of options
as a company).
The GPL is by far the most popular open source software license:
http://www.dwheeler.com/essays/gpl-compatible.html
The above-linked article makes the excellent point that what
potential developers think of a license is incredibly important
in deciding on a license -- if you want to recruit developers to
your project and you don't choose the GPL, the observed fact
is that it is harder.
Another argument for GPL instead of MIT/BSD for SAGE is
that -- well -- I don't trust the three companies Maple, Magma,
and Mathematica. I -- like *many* other SAGE developers --
are genuinely concerned that if SAGE were
a 100% BSD licensed project, many of its best parts would be
included into Maple, Magma, and/or Mathematica, and absolutely
nothing would be given back. For example, Magma currently
includes the LGPL licensed programs: GMP-ECM, GMP, MPC, MPFR.
They include GMP in Magma, do some things GMP does but
twice as fast, give nothing back, and post about the results
practically taunting the GMP developers. See section 3 of
this paper:
http://www.loria.fr/~gaudry/publis/fft.pdf
The GMP developers say on their website "Some well-known
non-free algebra systems are now using GMP. While this is a
recognition of GMP, these programs' non-free nature is unfortunate."
which suggests they aren't enthusiastic about this. It's
ridiculous to think that if GMP were BSD licensed, that Magma
would improve on it and give back their changes, since even
with an LGPL license, they don't give back improvements.
-- William
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-devel@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/
-~----------~----~----~----~------~----~------~--~---
