Environment name returned for Conda environments that do not have a name but only have a Path

[DonJayamanne]

When creating Conda enviornments using the Python Exension UI, the conda environment does not have a name.
However the Python extesnion API returns a name for the environment when it should not. In the past Python exension did not do this.
This breaks the code that is used to activate conda environments and install Python packages in Jupyter extension.

here's the work flow:

• Create conda env using Python extension
• The folder is .conda
• The environment msut be activated using the path and not a name as there's no conda enviornent named .conda
• The Jupyter extension now calls the Python Ext API to get the details of the Python exesnion API
  • The Python extension returns the details and the name of the envionment is .conda
• The jupyter extension now uses this to install python packages and uses the -n argument for conda and this falls over.

If you were to install the older version of the Python extnesion, now things work.
I.e. this is a breaking change in the Python exension

[DonJayamanne]

This is a regression in Stable vs Pre-Rlease.
if you check thge stable verison of Python extension, the envName is not returned

[DonJayamanne]

@karrtikr I'm hoping we can get this addressed before the release, as this introduces a regression at our end
As as metnioend the previous API didn't work this way.

[karrtikr]

Yes, I'm investigating the cause. The issue seems to be that when a .conda gets created, conda env list takes some time to return that environment. And if it doesn't return the conda environment, we assume it's a named environment by default.

It didn't happen earlier because discovery used to be slower, so by the time we got to running conda env list, the environment was already there.

So maybe the fix is to make it slower :P (Reverting the PR)

[DonJayamanne]

And if it doesn't return the conda environment, we assume it's a named environment by default.

From what I know you can look at the environments.txt file as well and if that doesn't contain an env name, then you can use env path.
Hence that's still very fast, faster than launching conda.

[karrtikr]

And if it doesn't return the conda environment, we assume it's a named environment by default.

Will likely change the default to assume it's a path based env, as -p works with everything.

environments.txt file as well and if that doesn't contain an env name

environments.txt might also not be updated in time when we look for the env. Also it never contains the env name, so not sure how that would help.

[karthiknadig]

@karrtik FYI. Conda env name computation is based on envs_dirs from conda info --json. If the patent directory is not one of the envs_dirs then it is not a named environment.
https://github.com/conda/conda/blob/fc92ea3353cce71df24fac31ce333ec9f8cf7ba7/conda/cli/common.py#L223-L224

I think at the time of resolving a conda environment we already have this info.