feat(pipenv): Use pipenv to lock down python dependencies

[jimexist]

this is really a trivial change and so far I haven't been able to find time to adjust the docker file accordingly.

This change is 

[k8s-ci-robot]

Hi @jimexist. Thanks for your PR.

I'm waiting for a kubernetes or tensorflow member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[coveralls]

Coverage increased (+0.1%) to 37.915% when pulling d4d5c04 on Jimexist:use-pipenv into a704246 on tensorflow:master.

[jlewi]

Thanks. I'm not familiar with pipenv so a couple questions

• How does this relate to requirements.txt? Can we get rid of requirements.txt?

• How will we use pipenv? Will pipenv be used to produce a pip package that will be used to install the py package in our containers?

[jlewi]

/ok-to-test

[jimexist]

How does this relate to requirements.txt? Can we get rid of requirements.txt?

yes. and you can always use pipenv to re-generate one

How will we use pipenv? Will pipenv be used to produce a pip package that will be used to install the py package in our containers?

for docker installation I can think of a template docker file like this (hope this can explain better):

FROM python:3.6

COPY Pipfile Pipfile.lock /my/app/path/

WORKDIR /my/app/path/

# see https://github.com/pypa/pipenv/blob/master/docs/advanced.rst#-deploying-system-dependencies
RUN pip3 install pipenv && pipenv install --system --deploy

COPY ./python_code /my/app/path/

CMD ["python", "script.py"]

[jimexist]

@jlewi now that you mentioned this - I found out that there are actually two use cases for python, one for test-infra (the airflow docker image) and one for releases (the /py/ folder).

the use case for pipenv could be... well, managing two separately. I can thinking this might sound like an overkill, but I'm kind of used to pipenv and hence the PR. Feel free to close it if you think it is too much. Personally I think it is a seamlessly working tool.

[jlewi]

According to this pipenv is the officially recommended package manager so I think it makes sense to use it.

Should we delete requirements.txt and update the developer_guide to refer to pipenv?

[coveralls]

Coverage increased (+0.08%) to 28.601% when pulling 34df25c on Jimexist:use-pipenv into ed7cae7 on tensorflow:master.

[coveralls]

Coverage remained the same at 28.518% when pulling 34df25c on Jimexist:use-pipenv into ed7cae7 on tensorflow:master.

[coveralls]

Coverage remained the same at 28.518% when pulling 34df25c on Jimexist:use-pipenv into ed7cae7 on tensorflow:master.

[jimexist]

@jlewi sure, WIP.

[jimexist]

@jlewi developer guide updated, PTAL

[coveralls]

Coverage decreased (-0.1%) to 28.416% when pulling 0b8f20f on Jimexist:use-pipenv into ed7cae7 on tensorflow:master.

[jlewi]

LGTM but looks like the build is still broken because of lint issues.

[jlewi]

@jimexist build should be fixed by #257 so I think you just need to merge in the most recent changes and push.

[coveralls]

Coverage remained the same at 28.458% when pulling 159623f on Jimexist:use-pipenv into 2793089 on tensorflow:master.

[jimexist]

@jlewi for some reason the build failed

[jlewi]

/test-all

[jlewi]

/test all