Should cognoml reside in a separate repo?

[dhimmel]

cognoml is the Python package we're developing for machine-learning

I'm starting to think this would simplify things.

@awm33 what do you think?

[awm33]

It might. So this repo would be the "playground", while cognoml is "production ready" and can be consumed by applications?

[dhimmel]

So this repo would be the "playground", while cognoml is "production ready" and can be consumed by applications?

Yeah, machine-learning will be a data analysis repository (tracks notebooks, figures, datasets). cognoml will be a python package repository. This will avoid confusion where it's not clear whether a file is needed for the package or for the exploratory analyses.

[awm33]

That makes sense

[dhimmel]

Okay, I'll create a cognoml repo once #62 is merged.

From this StackOverflow answer, here's a beastly git command to take only the package-relevant files and rewrite history to include only relevant commits.

git filter-branch \
    --prune-empty \
    --index-filter '
        git ls-tree -z -r --name-only --full-tree $GIT_COMMIT \
        | grep --null-data --invert-match "^LICENSE.md$" \
        | grep --null-data --invert-match "^setup.py$" \
        | grep --null-data --invert-match "^cognoml" \
        | xargs --null --no-run-if-empty git rm --cached -r
    ' \
    -- \
    --all

This command currently will transfer the LICENSE.md and setup.py files to the new repo as well as everything in the cognoml directory.

[dhimmel]

Okay I used the above command to populate the new cognoml repository up to cognoma/cognoml@d4ab12c.