Re: [Help-gsl] Spearman rank correlation coefficient

[Rhys Ulerich <rhys dot ulerich at gmail dot com>]

Hi Tim,

> a month ago I proposed an implementation of the Spearman rank
> correlation coefficient ...
> https://gist.github.com/1784199#file_spearman_v2.c
>
> ... Thus can I submit it to the GSL main trunk?
> I've never done that before. Can someone indicate me what to do?
> Should I request "developer write access" for instance?

>From reviewing the code at the github...

Your gsl_stats_spearman_workspace (e.g.) should be passed into your
gsl_stats_spearman method.
You'll want to declare a struct-based workspace rather than
allocate/deallocate many separate variables in your
gsl_stats_spearman_alloc/gsl_stats_spearman_free.  As an example,
check out gsl_bspline_workspace from bspline/gsl_bspline.h which is
allocated/deallocated by gsl_bspline_alloc/gsl_bspline_free.  You'll
want to add the allocation-failure error handling as in the bspline
example.

You gsl_stats_spearman_rank implementation should always be checking
if calloc/realloc failed and handle that error appropriately

Helper functions like "resolveTies" which should not be publicly
visible should be declared static.  That way they can only be called
from within the same file.

Finally, the Apache License 2.0 origin of the logic should be made
more clear in the file.  Just attributing the logic with "Source"
seems insufficient-- I'd organize the file so everything below some
point if the Apache-derived code and everything above that point is
your GSL-ready GPL3 logic.

Give a shout when you're ready and I'll take another look.  I can help
with getting the final logic into GSL's trunk.

- Rhys