Incomplete implementations?

[jtappin]

A question for the developers.

Is there any point in contributing/adding library routines that don't include all of the features in the version in the IDL routine?
Specifically I have an implementation of r_correlate that does not include the optional outputs PROBD and ZD as (a) I didn't need them and (b) I'm not sure how to implement them.

It needs a bit of work to integrate the Kendall coefficient which is currently a completely separate routine. It is actually much faster than IDL does it, and also gives correct answers for larger datasets.

[alaingdl]

Any contribution is welcome as long as it can be made under the GNU GPL licence. Some tests cases are also very welcome.

I would say incomplete code is not a problem, as long as the missing part is clearly visible
(clear warning/error messages). In the past I used to have interns to complete some code and missing functionalities ..