[ODE] ODE's license

nlin@nlin.net nlin at nlin.net
Mon Jun 24 11:50:02 2002


> this change will probably happen in a few days. the (very few) people
> who have contributed code to the ODE core may chose to object to this,
> as they may have expected that their code would be licensed under LGPL,

No objection from me - somewhat ironically, my MSVC DLL changes were made
in order to enable easier compliance the LGPL, but I can understand how
users in commercial situations might have contractual difficulties with
their clients and the LGPL.
> 
> also: various people have contributed code to ODE that i've been sitting
> on for a while. i'll get around to this eventually, but i've had very
> little time for ODE recently. one thing i'll consider is hosting ODE
> on sourceforge and allowing any interested parties to be administrators
> of the project (and thus you can add your own code - but you have to
> test and document it yourselves as well :-) ).

Personally I kind of like the current situation where I can be fairly
certain that the person who probably knows the code best (Russ) will
manually integrate changes rather than having a wild-west free-for-all
with the source code. However if someone has contributed a bleeding edge
feature that you desperately need, I can also understand that the slow
release schedule might be hampering.

My suggestion would be the addition of an "incoming" directory where
not-yet-integrated patches are simply dumped but can still be checked
out vis CVS. Those absolutely needing bleeding-edge functionality can
hack this into their own code; when Russ gets around to integrating it,
it will be deleted from "incoming" and become part of the official release.

This of course requires that Russ have enough time to eventually
integrate the patches, but as I said, I would feel most comfortable
with that solution.

If a "free-for-all" CVS code repository is opened up, then I strongly
suggest that as a first step Russ publish a detailed reference list to
the particular research papers (Anitescu, Stewart, Potra, etc) used as
a theoretical basis for ODE, plus document any changes to or specializations
of these techniques. For instance, there appears to be a lot of speculation
and possibly misunderstanding about the details of the time-stepping
integration scheme used in ODE, and I would be wary of any changes
to this part of the code without first being able to refer to a detailed
theoretical explanation of how it is supposed to work. 

Another random suggestion, would it be possible to set up an ODE Wiki
on q12.org? This way users and programmers could contribute documentation
easily. Hopefully this will expand into a self-sustaining and organized
body of knowledge about ODE (in particular, ODE internals).

My 0.02,
-Norman