Conda recipe for bloscpack

[ijstokes]

recipe requires cram conda package (-c ijstokes conda channel) and currently comments out coveralls -- but otherwise builds

[coveralls]

Coverage remained the same at 95.303% when pulling 476aa468232a845a388e9f8196a5cebb602fed2a on ijstokes:ijstokes/conda_recipe into f9e528318004eb974b4d2cf1c2ea7cc463a9fc5d on Blosc:master.

[FrancescAlted]

Thanks for that, but I think it would be better if we keep packaging config files out of the repo. Thoughts?

[esc]

I don't think it matters much anymore. For python-blosc we did that:

https://github.com/Blosc/python-blosc-conda-recipe

I think I even made some binaries that I uploaded to the Blosc channel on anaconda.org:

https://anaconda.org/blosc

But on anaconda.org everyone just seems to do whatever they like anyway and there is a plethora of blosc packages from all sorts of different people:

https://anaconda.org/search?q=Blosc

So in the end, unless someone can provide technical arguments for where the conda packaging stuff should go, I really don't care.

[esc]

We also did this for bcolz:

https://github.com/Blosc/bcolz-conda-recipe

But I don't think that recipe ever worked. In fact it seems to live in the repo too:

https://github.com/Blosc/bcolz/tree/master/conda.recipe

So yeah I am not sure.

From a continuous build/integration/deployment perspective, having the recipe in the repository would make it easier to build new development releases on travis and upload them automatically to PyPi and anaconda.org. That might be a nice feature to add as we transition into an automated continuous paradigm.

[esc]

The failed test is due to an instability in the test, I am fixing it now.

[esc]

Also, I would suggest to merge this for now. If it helps @ijstokes .

[esc]

I restarted the command-line tests and everything is fine now. Also a fix for the test instability is here:

https://github.com/Blosc/bloscpack/pull/49

[esc]

Will merge this now, since it is green.