Hi Emmanuelle [This in reply to the discussion from https://github.com/scikits-image/scikits.image/pull/19 ] On Fri, Sep 16, 2011 at 11:20 AM, emmanuelle <reply+i-1613972-89d62c860f1868f0cf5f68ffc7b222a2e568a670@reply.github.com> wrote:
1. Yes, it seems a good idea to centralize these files in a utility module. In the future, maybe we can keep only one implementation of heaps, either the one by Almar Klein/Zach Pincus, or the cellprofiler one. But it would take some time to refactor this code, and the files that depend on these modules, so we'd better keep that for the next release. In the meantime, we can use a util module to store these files.
Agreed. I'll all move them to the same place for now (maybe _util to discourage their use by users for the time being), but we'll leave the refactoring until later.
2. I don't really see the point in separating the cellprofiler code in a submodule. For example, cpmorphology contains a lot of core routines (skeletonization for example) that should be directly under morphology.
I don't think I expressed my idea clearly. I just wanted to move all the .pyx files etc. to a sub-directory of morphology, and them import them in the __init__ into the morphology namespace. The only reason for that was because the cp code had so many files--I thought it might make it difficult to follow the morphology code if we didn't separate it in some way.
3. I was thinking of stealing the Scikit learn config for generating Sphinx documentation from python example files (+ their way to generate a gallery). It doesn't seem to hard to reuse.
That would be great. The current tutorial system is cumbersome. I'd like something fairly simple and transparent, and if we can borrow it from them so much the better.
I can help you with that, but not this week-end! Next week I can find some time. When do you plan to do the release?
I'm having a mini-sprint at Stanford today and will be working all day Sunday on completing the release. Realistically, I won't get everything done on Sunday, but we should be much closer (I'll try and merge as many PR as possible). The week thereafter, we can round things up. Due to time zone differences, we probably won't all be able to meet at the same time, but let's coordinate so that, by latest, we release next weekend (24, 25 September). Regards Stéfan