Test finalized API for python environments

[karrtikr]

Refs: #19101

• anyOS: @IanMatthewHuff
• anyOS: @DonJayamanne

Complexity: 5

Create Issue

---

Requirements

Pre-assigning Jupyter team, but note builds are not available yet, testing begins tomorrow.

1. Use Pre release build of the python extension
2. Have as many types of environments as you can.
3. Interesting types to test: Conda, Pyenv global installs, Pyenv conda envs etc.

Test APIs exposed via ProposedExtensionAPI

• See ProposedExtensionAPI for APIs to test: https://github.dev/microsoft/vscode-python/blob/main/src/client/proposedApiTypes.ts
• Use Python: Clear cache and reload command to clear any pre-existing discovery cache in the extension.
• Try dry-adopting it in the Jupyter extension for discovery of Python environments and make sure it covers all scenarios.

Example usage:
Copy over contents of https://github.com/microsoft/vscode-python/blob/main/src/client/proposedApiTypes.ts#L8 as needed.

const extension = extensions.getExtension('ms-python.python');
if (extension) {
    if (!extension.isActive) {
        await extension.activate();
    }
    const api: IExtensionApi & ProposedExtensionAPI = extension.exports as IExtensionApi & ProposedExtensionAPI;
    if (api.environment) {
        const envPath =  api.environment.getActiveEnvironmentPath();
    }
}

[IanMatthewHuff]

Right now I'm testing this with a separate extension as opposed to trying to fully adopt into Jupyter. Seemed like a fairer test of the API itself.

• After refresh list of envs looks correct in .all
• getActiveEnvironmentPath returns the correct Env, as well as after a change
• After calling resolveEnvironment the env looks correct (venv, conda, global, pyenv)
• updateActiveEnvironmentPath updates the environment
• onDidChangeActiveEnvironmentPath fires corrected with the right arg (and doesn't file if update parameter is already active)
• onDidChangeEnvironments changes when adding a new Env
• Cancellation token works on a full force refresh

[karrtikr]

Yep, by dry-adopting I mean we should make sure it covers all scenarios in the Jupyter extension.

[IanMatthewHuff]

Yep, by dry-adopting I mean we should make sure it covers all scenarios in the Jupyter extension.

That seems a bit much for a single TPI. But so far things all look good with it. Maybe Don knows better if we can just drop this in quickly.

[DonJayamanne]

i agree with Ian, it's impossible to cover ALL scenrios in a TPI without adopting this in jupyyer extension

if you're plan is to finalize the api based on feedback from jupyter extension adoption, then i don't think we can finalize this api now.
All we (jupyter extension) can do is test this with a simple sample and provide some feedback.

we will discuss this in our stand-up.

[karrtikr]

I see, in that case just testing and feedback would suffice for purposes of this TPI.

Will discuss with the team on when we plan to finalize it.

[karrtikr]

So I talked with the team and decided that we'll finalize it in an iteration or two, hope that gives Jupyter some time to adopt the API. Meanwhile we'll still announce it in our blog posts and release notes to get additional feedback!