[Nipy-devel] Fwd: 1st/2nd level analysis modules in python
Fri Nov 6 15:34:25 CST 2009
> i would like to derive the volumetric roi class from the base image
> class in nipy, as it shares several common properties. the last
> discussion on the list still evolved around dimensions, their naming
> and the conventions of coordinate systems. this didn't effuse a whole
> sense of stability of core code.
It's important to get beyond 'sense of stability' to concrete
questions like 'can I rely on interface X being supported in the near
term'. Can you specify your worry in that way?
> the other aspect that i would really like to use nipy for is
> model/design specification. and i would like that api to be reasonably
> stable as well.
The best way for that to happen - in this open-source world - is to
pitch into trying to work with and improve what we have.
> i would like to know whether nipy is seen as something analogous to
> vtk/itk? is there a core set of data structures and algorithms that
> forms the api? while some of these vtk/itk algorithms and their api
> have evolved for these other projects over time, their data structures
> have been fairly stable over extended periods. is this something i can
> expect from nipy?
In due course, but these projects took many years of gestation. If
you want that to happen quicker, you need to contribute.
> while i personally don't have a problem with a BSD license, i have
> spoken with some developers who do. they would either like to see a
> dual licensing scheme (GPL,commercial) before they would consider
> contributing to the project. they have no problems with free
> academic/research use, but aren't willing to see a company make a
> profit with their hard work without contributing in some significant
> way to the project.
I know, this discussion generates some strong gut feelings. I
personally have no problem with a company making a profit with my
work, as long as the project benefits. In practice I think that's
what the BSD gives - the greatest chance of benefit,
More information about the Nipy-devel