Currently there's a private API that Pylance gets to use in order to determine the Python Env associated with Jupyter notebooks.
Unfortunately there are a few problems with this API:
Its built specifically for Pylance, other extensions cannot get this same information
Upon selecting a new Kernel, the Pylance extension won't always know that a new Python Env has been selected for a Jupyter Notebook
The API is inconsistent, currently there's an API in Python extension to get the Python Env associated with a Uri, but Jupyter Notebooks is not part of this (when Jupyter Notebooks are also just Uris).
Pros:
Unified approach to getting Python Environments for notebooks, Python files, workspaces etc.
No code change for extensions consuming the Python API
Resolves some bugs as well
Solution:
Keep the Python extension + Jupyter Extension API internal (as today)
Update the Python extension as follows:
export type ActiveEnvironmentPathChangeEvent = EnvironmentPath & {
/**
* Workspace folder the environment changed for.
* Or Resource the environment changed for.
*/
readonly resource: WorkspaceFolder | Uri | undefined;
};
Internally, the call to getActiveEnvironmentPath(resource?: Resource): EnvironmentPath; would now handle Uris a little differently. If the Uri is a notebook Uri, then get the interpreter from Jupyter extension (using existing code).
The check if the Uri is a notebook can optionally be passed into Jupyter as well, i.e. keep it super simple.
Currently there's a private API that Pylance gets to use in order to determine the Python Env associated with Jupyter notebooks. Unfortunately there are a few problems with this API:
Pros:
Solution:
Internally, the call to
getActiveEnvironmentPath(resource?: Resource): EnvironmentPath;would now handle Uris a little differently. If theUriis a notebook Uri, then get the interpreter from Jupyter extension (using existing code). The check if the Uri is a notebook can optionally be passed into Jupyter as well, i.e. keep it super simple.