Conda Build v3 revamp

[jaimergp]

Following updates in conda-dev-recipes, I will start adapting (very conservatively) conda-recipes to the new CB3 infrastructure running on Azure.

To do

• Azure gets triggered
• Recipes build in:
  • Linux
  • MacOS
  • Windows
• All recipes are compatible with the new infrastructure (no extra fields are needed)
• Uploads work correctly
• Decide whether to bring openmm from conda-recipes-cuda or update that one.
• Identify packages eligible for noarch: python

Packages

Builds - :white_check_mark:
Does not build - :warning:
Fix needed - :hammer_and_wrench:
To be removed - :wastebasket:
Migrate to CF - :fast_forward:

Package	Builds?	Keep/remove/migrate
ambermini-16.16.0-7.tar.bz2	⚠️	🗑️
cclib-1.5-py37_0.tar.bz2 -> (CF)	✅	🗑️
cerberus-1.1-py_1.tar.bz2 -> (CF)	⚠️	🗑️
clusterutils-0.3.1-py37_1.tar.bz2	✅	⏩
cmiles-0.1.5-py37_2.tar.bz2	⚠️	⏩
covar-0.2-py37_0.tar.bz2	⚠️	⏩
cpptraj-2016.0.07-0.tar.bz2	⚠️	🗑️ ?
deepchem	???	⏩
eigen3-3.3.7-0.tar.bz2 -> (CF eigen)	⚠️	🗑️
fahmunge-0.3.0-py37_0.tar.bz2	✅	⏩
fftw3f-3.3.4-2.tar.bz2 -> (CF fftw)	⚠️	🗑️
forcebalance-1.7.0-py37_1.tar.bz2	⚠️	⏩
fragmenter-0.0.4-py37_0.tar.bz2	⚠️
iapetus-0.1.0-py37_0.tar.bz2	✅
jom -> (CF)	???	🗑️
lomap	???
m2r-0.1.15-py37_0.tar.bz2 -> (CF)	✅	🗑️
mdentropy-0.3.0-py37_0.tar.bz2	⚠️	⏩
molpx-0.1.5-py37_0.tar.bz2 -> (CF)	✅	🗑️
msmb_data-2.0.0-0.tar.bz2	✅	⏩
msmexplorer-1.1.0-py37_0.tar.bz2	⚠️	⏩
munkres-1.0.7-py37_0.tar.bz2 -> (CF)	✅	🗑️
omnia-1.0-1.tar.bz2	✅
openforcefield-0.6.0-py37_0.tar.bz2	⚠️
openforcefields-1.0.0-py37_0.tar.bz2	✅	⏩ ?
openmm-setup-1.1.1-py37_1.tar.bz2	✅
openmmtools-0.19.0-py37_0.tar.bz2	⚠️
openmoltools-0.8.4-py37_1.tar.bz2	✅
openpathsampling-1.0-py37_0.tar.bz2	✅
oset-0.1.3-py37_1.tar.bz2 -> (CF)	⚠️	🗑️
osprey-1.1.0-py37_0.tar.bz2	✅	⏩
packmol-1!18.013-0.tar.bz2 -> (CF)	⚠️	🗑️
parmed-3.2.0-py37_0.tar.bz2 -> (CF)	⚠️	🗑️
pdbfixer-1.6-py37_0.tar.bz2	⚠️
perses-0.4.0-py37_0.tar.bz2	⚠️
propertyestimator-0.0.9-py37_0.tar.bz2	⚠️
pybtex-docutils-0.2.1-py37_1.tar.bz2 -> (CF)	⚠️	🗑️
pytest-ipynb	???
pygraphviz-1.3.1-py37_0.tar.bz2 -> (CF)	⚠️	🗑️
quadprog-0.1.6-py37_0.tar.bz2	⚠️	⏩
saltswap-0.5.2-py37_0.tar.bz2	✅
schema-0.6.2-py37_0.tar.bz2 -> (CF)	✅	🗑️
scripttest-1.3-py37_0.tar.bz2 -> (CF)	✅	🗑️
shiftx2	?????	⏩
silo-4.10.2-0.tar.bz2	⚠️	⏩
smarty-0.1.5-py37_0.tar.bz2	✅	⏩
smirnoff99frosst-1.1.0-py37_1.tar.bz2	✅	⏩
solvationtoolkit-0.4.3-py37_0.tar.bz2	✅
sphinxcontrib-lunrsearch-0.3-py37_0.tar.bz2	⚠️	⏩ ?
svgwrite-1.1.6-py37_0.tar.bz2 -> (CF)	✅	🗑️
yank-0.25.2-py37_1.tar.bz2	⚠️
yank-examples-1.1.0-0.tar.bz2	✅

Changelog

• 2020.02.03 - Attempt to classify recipes into deprecated or conda-forge ready. First group has its own directory (deprecated); CF-ready are identified by a CONDA_FORGE_VALID file in their recipe folder.
• 2020.01.27 - Copied scripts from conda-dev-recipes (with uploads disabled), deleted Travis and Appveyor configs.

[jaimergp]

Added a summary of the build status of the different recipes.

I think we should also take some time to decide which packages are going to "make it" to the CB3 revamp (or optimistically to conda-forge). I guess that those who won't make it it's because:

• they were dependencies that are now available in CF.
• they are no longer being maintained / developed.

Can we choose which ones to keep and which ones to deprecate? @jchodera @Lnaden

[jchodera]

My understanding is that we have already removed recipes for packages available on conda-forge, and that remaining ones likely need to be migrated. Some may no longer be actively maintained, however.

What's the best way to mark what needs to transition and what should be deprecated? Should we mark this in the yaml files or in this PR?

Also, should we add a list of maintainers to the yaml file, or will it be understood the recipes left to maintain are up to us?

[jaimergp]

I have moved those already in CF to deprecated. Then added a CONDA_FORGE_VALID blank file in those remaining under recipes if they don't seem to have an incompatible dependency.

[jaimergp]

I'd like to include extra.maintainers too, wishing somebody else will volunteer to participate :) But yes, at first I see us as the only maintainers.