Use the json API on pypi to get descriptions for the pypi packages

[ericdill]

cc @tonyfast

http://stackoverflow.com/questions/21419009/json-api-for-pypi-how-to-list-packages

It would be great to include the description of the package in the package requirements in the meta.yaml of the conda-recipe

e.g.,

• https://pypi.python.org/pypi/scikit-image/json
• https://pypi.python.org/pypi/scikit-xray/json

[tonyfast]

I'll try to see if I can add something this, but just so we have it. Here is the Conda package metadata

[ericdill]

Thanks for the continuum package info link! With 8157025 in #10 , the info from continuum has been incorporated into the meta.yaml generation. It is configurable which order to search in (i.e., pypi then continuum or continuum then pypi).

The meta.yaml now looks like this:

requirements:
  build:
    - python
    - numpy # (conda) array processing for numbers, strings, records, and objects.
    - setuptools # (conda) Easily download, build, install, upgrade, and uninstall Python packages
    - six # (conda) Python 2 and 3 compatibility utilities
  run:
    - python
    - lmfit # (pypi) Least-Squares Minimization with Bounds and Constraints
    - netcdf4 # (conda) python/numpy interface to netCDF library
    - numpy # (conda) array processing for numbers, strings, records, and objects.
    - scipy # (conda) Scientific Library for Python
    - six # (conda) Python 2 and 3 compatibility utilities
    - xraylib # No package info found for xraylib at any of these locations: ['conda', 'pypi']

test:
  requires:
    - nose # (conda) nose extends unittest to make testing easier
    - scikit-image # (conda) Image processing routines for SciPy

[tonyfast]

I think if no information is found then the comment should be left blank.

Can conda build from a github repo like pypi can? This syntax ... - git+https://github.com/ericdill/conda-skeletor.git

[tonyfast]

Also, it might be preferable to put the package distribution at the end of the string.

requirements:
  build:
    - python
    - numpy # array processing for numbers, strings, records, and objects. (conda) 
    - setuptools # Easily download, build, install, upgrade, and uninstall Python packages (conda) 
    - six # Python 2 and 3 compatibility utilities (conda) 
  run:
    - python
    - lmfit # Least-Squares Minimization with Bounds and Constraints (pypi) 
    - netcdf4 # python/numpy interface to netCDF library (conda)
    - numpy # array processing for numbers, strings, records, and objects. (conda) 
    - scipy # Scientific Library for Python (conda) 
    - six # Python 2 and 3 compatibility utilities (conda) 
    - xraylib 

test:
  requires:
    - nose # nose extends unittest to make testing easier (conda)
    - scikit-image # Image processing routines for SciPy (conda) 

[ericdill]

Yeah that seems pretty reasonable

[ericdill]

I'm not sure what you mean by

Can conda build from a github repo like pypi can? This syntax ... - git+https://github.com/ericdill/conda-skeletor.git

Is this a pip install <git_url> thing?

[ericdill]

Like --> pip install https://github.com/NSLS-II/metadatastore/zipball/master#egg=metadatastore

[tonyfast]

This is what I was referring to.

This way you can pip install using shorthand to a version control system.

[ericdill]

I have not seen a conda equivalent of pip install -e <path_to_git_repo>