[SciPy-dev] how to help scipy development?

Pearu Peterson pearu at scipy.org
Wed Nov 16 14:21:33 CST 2005



On Wed, 16 Nov 2005, Arnd Baecker wrote:

> Nils' question on how to write unit tests for odeint
> raises an important point:
> How to encourage contributions to scipy?
>
> My impression is that somehow the gap between "end-users"
> and developers should be made smaller.
> For this it might be helpful to have (detailed)
> instructions on:
> - which information to provide for bug reports
>  (a script which generates all the info would be great:
>  scipy.__buginfo(fname="bug.txt")
>  A lot of this is already available via
>    from f2py2e import diagnose
>    diagnose.run()
>  )
> - how to provide useful information when segfaults occur
>  (invocation of gdb ...)
> - how to add documentation to routines and how to submit these
> - how to write unittests
> - how to add additional wrappers to lapack (etc.)
> - how to provide patches
> - how to add new tools to scipy
>  (see core/scipy/doc/DISTUTILS.txt
>   and scipy/DEVELOPERS.txt)
>
> Some of these points might seem trivial to expert developers,
> but can be a significant barrier for newcomers.

I can write docs for most of the above howtos. In fact, asking such 
well-defined questions is a very good start to write docs.

However, the open question is still where to put these howtos so that 
users will find them easily and also so that it would be easy to maintain 
them. core/scipy/doc/ would be one place to put rest-docs.

> In order to not completely bankrupt my opinion/suggestion budget  ;-)
> I have added some remarks in the other thread on how one could
> set up unittests (hope I got that right ...).

Thanks! I have few notes on your remarks though, I'll make them in the 
other thread.

Pearu




More information about the Scipy-dev mailing list