[SciPy-Dev] Procedure for new code submission (pyHHT)

[Pauli Virtanen]

Stuart Mumford <stuart <at> mumford.me.uk> writes:
> Hello,I am working on a EMD / HHT code for some analysis 
> I am doing as part of my PhD. (It is currently here:
> https://github.com/Cadair/pyhht (kicked off from jaidevd's inital work))
> I was wondering what the general procedure for getting something
> like this included in SciPy is. I know it requires a lot more work
> to even be considered but I would like to focus what I am doing
> with that end in mind.

If the code is something that is useful across scientific fields, then IMO
it can be bundled with Scipy.

In this case, we have the misfortune of knowing beforehand
that the algorithm is patented in US, and although as ridiculous
it is, this makes the code poisonous for many US-based users.
So, I'd say that this cannot be included in Scipy which does
have US-based users and redistributors...

> The Web here:
> http://scipy.github.com/dev-zone.html says that initially you should
> write a SciKit, is this still the way to go or is it fine just to
> create a repo with good quality documentation etc? 

The information on that page is a bit outdated. This one is more up to date:

    https://github.com/scipy/scipy/blob/master/HACKING.rst.txt

Note that being a "SciKit" is mostly just a matter of branding --- you make a  
separate Python package from your code as usual (which you seem to be working 
towards), call it "PyHHT Scikit", and add it to the "Topical software" list at 
scipy.org

We earlier recommended naming the package as "scikits.XXX" but this
turned out to bring in some technical problems in the long run, so
that part is optional.

> Also is a HHT toolbox something that should be integrated
> into scipy.signal or (at risk of restarting an amusing debate)
> should it go into scikit.signal??

Without the patent issues, scipy.signal would be the correct location.
I'm not sure if scikit.signal is dead or just sleeping.

-- 
Pauli Virtanen