refreshEnvironment does not return the latest details of the conda environment

[DonJayamanne]

Based on discussions with @brettcannon
Such an API will be provided that allows Jupyter ext to install Python into an empty conda env

Here's what is happening

• Jupyter installs python into the conda env
• Jupyter ext uses the resolveEnviornment API to get the latest conda env info
• Unfortunately the latest conda env info is not yet returned by the Python extension (Python Ext returns the old info which leads one to believe that Python is not installed into the conda env)

[brettcannon]

Added to the iteration plan.

[DonJayamanne]

As part of this API call please can we ensure the Conda environment is updated
I.e. ensure the change event is triggered to update the conda env information, such as updating the Pytohn version, Uri (path to the executable etc)

Today I have a PR that installs Python into the conda environment from within the Jupyter extension (via custom code), however refreshing the Conda env takes more than 1 minute, hence that approach is not the best

i.e. i will need the Python Extension api to

• Provide a way to install Python into a conda env
• As well as update the conda env properties after the installation (so that environments contains the right information for this conda env)

[brettcannon]

ensure the change event is triggered to update the conda env information, such as updating the Pytohn version, Uri (path to the executable etc)

Today I have a PR that installs Python into the conda environment from within the Jupyter extension (via custom code), however refreshing the Conda env takes more than 1 minute, hence that approach is not the best

We can trigger an update immediately after the install, but I don't know if it will necessarily lead to a faster outcome since conda chooses what Python version to install.

[DonJayamanne]

faster outcome since conda chooses what Python version to install.

sorry, what I meant was the update of the environment information (environment info in api) would be faster if python extension updated the information for that specific conda env immediately
As opposed to Jupyter extension calling the refresh api (which takes more than a minute for me on my Mac)

[brettcannon]

sorry, what I meant was the update of the environment information (environment info in api) would be faster if python extension updated the information for that specific conda env immediately

Ah, so you're asking to just make sure that a changed event fires with the updated information once conda can provide that info to us instead of you having to to a manual trigger?

[DonJayamanne]

Yes thats correct, thanks

[brettcannon]

We were talking about this API in stand-up today, and we were wondering if you are calling conda install ipykernel? If you are then does conda itself not automatically trigger the Python installation as a dependency for ipykernel since conda supports multiple programming languages? If that's the case then would it be better to let conda handle the installation of Python implicitly as part of your installation call so it can make sure that a version of Python that's compatible with ipykernel gets installed?

If conda doesn't trigger an installation of Python then we should file a bug on conda to fix this. And if you're not running any installation command then this idea doesn't work. 😅

[DonJayamanne]

Will try that out and let you know