Making the package installable with pip and conda

[tompollard]

Hi @kjohnsson, many thanks for sharing this work! Have you thought about making it available on pypi and conda-forge?

I'd like to make use of the tests in some code that I'm working on, and it would be easier to do this if the package could be installed with pip.

I am happy to help with refactoring, preparing the code for distribution, etc, so let me know if this would be of interest.

[kjohnsson]

Hi Tom! I thought I'd make it available on pypi eventually, maybe it can be sooner rather than later ;-) I planned to first do some improvements (e.g. using numba to increase speed of dip test and using something neater than MPI for running on multiple cores) - but maybe you'd prefer to have the version basically as it is if you can have it sooner? I haven't understood yet if pypi puts any requirements on packages (e.g. on documentation, tests et.c.), do you know what steps needs to be taken? Unfortunately I don't have much time for this project at the moment, so I'd appreciate any help.

[tompollard]

Great, thanks Kristen. If you're good with me contributing, I'll make a start on refactoring the code and will submit the changes as a series of pull requests. You can then review the changes and merge with the master branch if you are happy.

I can keep things simple to begin with, focusing on:

• tidying up the package for usability and distribution
• setting up the tests to run on a continuous integration server
• creating a Jupyter Notebook to demonstrate usage

Does this sounds okay on your side? If so, I'll make a start :)

[kjohnsson]

That sounds great! =)