Quick fix .gitpod.yml

[noklam]

NOTE: Kedro datasets are moving from kedro.extras.datasets to a separate kedro-datasets package in
kedro-plugins repository. Any changes to the dataset implementations
should be done by opening a pull request in that repository.

Description

Gitpod is broken when you open a new workspace. I observe at least 2 things are not working

1. Pre-build is not happening - everything get installed when you open a new workspace and take ~5-10minutes before you can do anything
2. the template project isn't created because of kedro is not found. This is likely linked to 1.

Development notes

The PR simply add one step to reload ~/.bashrc, it's unclear why the CLI isn't installed with pip, but this would work.

This PR only fix 2. as a quick hack. The pre-build stop working since GitPod move to "organisation", we create a new organisation and works for a few months but it started to fail again recently.

Checklist

• Read the contributing guidelines
• Opened this PR as a 'Draft Pull Request' if it is work-in-progress
• Updated the documentation to reflect the code changes
• Added a description of this change in the RELEASE.md file
• Added tests to cover my changes

[astrojuanlu]

I can confirm:

On main

Installing collected packages: kedro
Successfully installed kedro-0.18.8
bash: kedro: command not found

With this PR

| | _____  __| |_ __ ___
| |/ / _ \/ _` | '__/ _ \
|   <  __/ (_| | | | (_) |
|_|\_\___|\__,_|_|  \___/
v0.18.8

Kedro is a Python framework for
creating reproducible, maintainable
and modular data science code.

No plugins installed

For future work, I'd like to note that this flow looks a bit contrived, and I'm not sure what are the advantages over just doing pip install ..., similar to what I do in my workshop:

https://github.com/astrojuanlu/workshop-jupyter-kedro/blob/main/.gitpod.yml#L5-L7

and a custom Dockerfile:

https://github.com/astrojuanlu/workshop-jupyter-kedro/blob/main/.gitpod.Dockerfile

Bear in mind that pip install --user might start breaking soon too, see https://pythonspeed.com/articles/externally-managed-environment-pep-668/

[noklam]

I like the Docker approach, for the time being I am just annoyed that I have to fix the shell every single time.

[antonymilne]

Thank you for the fix!

[antonymilne]

Would be nice to add a comment here explaining why it's needed (even if it's not really clear why...)

[antonymilne]

@astrojuanlu the current approach is indeed very contrived, but there are some reasons for the madness.

kedro/.gitpod.yml

Lines 4 to 7 in 6df4d40

	# We want packages installed during the pre-build init steps to go to /workspace
	# rather than ~ so that they are persisted. Gitpod sets PIP_USER=yes to ensure this,
	# but pre-commit requires PIP_USER=no. Hence we set PIP_USER=no and use
	# pip install --user to install to /workspace.

Gitpod always spoke of persisting things outside /workspace after a pre-build, but last time I checked it was not possible still and so the workaround with --user was still needed. So unless something has changed here, I think you'll find that if you enabled pre-builds on https://github.com/astrojuanlu/workshop-jupyter-kedro/ your requirements would no longer still be available after the init stage.

A custom Dockerfile is definitely another way to do this, but it's awkward to pip install the right set of requirements because the repo isn't checked out in time for us to look at requirements.txt. See https://stackoverflow.com/a/70416361.