[SciPy-Dev] documenting fortran wrapped functions, was:Problems with documentation of scipy.interpolate.dfitpack
Charles R Harris
Tue Jul 13 09:23:25 CDT 2010
On Tue, Jul 13, 2010 at 8:15 AM, Ralf Gommers
> On Fri, Jul 9, 2010 at 3:33 PM, <email@example.com> wrote:
>> On Thu, Jul 8, 2010 at 1:18 AM, Vincent Davis <firstname.lastname@example.org>
>> > On Wed, Jul 7, 2010 at 10:09 PM, bowie_22 <email@example.com> wrote:
>> >> Hello together,
>> >> I tried to contribute to the documentation of scipy.
>> >> I found out that the whole
>> >> scipy.interpolate.dfitpack
>> >> package seems to be a little bit different.
>> >> It is not possible to view the code.
>> >> You get 404 File not found error.
>> >> Is there no source code available for dfitpack?
>> >> If so, how is the documentation handled?
>> > I think the first line in the doc editor says it all.
>> > "This module 'dfitpack' is auto-generated with f2py (version:2_8473).
>> > Functions:"
>> > I think this is the source
>> > Vincent
>> Many of the fortran based functions are underdocumented, e.g.
>> lapack/blas in linalg, most functions in special, often the fortran
>> files contain quite extensive documentation. But it seems in some
>> cases like dfitpack, some documentation are autogenerated.
>> Just a question because I don't have the time to look into this:
>> Do we need to mark f2py wrapped functions as "unimportant", or can we
>> make sure every f2py wrapped function has a editable docstring?
>> I looked at this a while ago after writing docs for ftpack_lite. The
> problem I ran into is that the add_newdoc function only works for C code,
> not Fortran code. Not sure if this can be easily fixed or not.
I've been thinking of redoing parts of that code in Cython. Not that there
looks to be a lot of time for that, but the basic functions are small and
the driver functions should be a lot cleaner with the base functions in c.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the SciPy-Dev