continuumio / pypi-conda-builds

build_from_pypi automatically builds Conda packages from PyPI. To build top n sorted by download count packages run:

python build_from_pypi.py -n 100 --recipe --build --pipbuild

This will first create recipes using conda-skeleton for the top 100 packages, then it uses conda-build to create Conda package out of those recipe. If this fails build_from_pypi will try to use the pipbuild process.

pypi-conda-builds's Issues

Handle hung build scripts

The erratically behaving build scripts can stall the whole build_from_pypi, check the amount of time they are taking if it is too much kill them and maybe restart with more time.

List Reason's for build failures from recipes created using conda-skeleton

• Build dependencies are not satisfied.
• conda-skeleton uses the wrong name for the packages. Example conda package for beautifulsoup4 is called beautiful-soup
• conda-skeleton converts all package names to lower case while the list of packages to be build uses the original case, hence if the original case is not all lower case it cannot find the recipe directory.

BinPy build fails

The build of BinPy fails. Having a look at the error log suggest that L52 in the meta.yaml is erroneous. Learn about the internals of conda skeleton and how it creates the import tests.

Sort the list of packages by number of downloads

Skip creating recipes for packages in `list_of_recipes` and `list_of_failed_recipes`

conda build needs to be in the current working directory to build the dependencies of the project.

Analyze the reason for failed recipes

The current create_recipies.sh script has tried to create for recipes for some thirteen thousand packages of which there are more than twenty-five hundred failures. Analyze the reasons for the failures and fix them. Few of the known reasons are:

• The package is non-existant
• There are more than one source version available and the script doesn't choose between them.

Write Blog Post

It should contain:

• Overview of Conda, Conda-build and Conda-skeleton
• Resource used to learn shell scripting
• Things I liked about shell scripts
• PyPi and download stats
• AWS
• screen