Conda environments

Create or refresh an environment

We call our environment std, for ‘standard’.

1. If an environment of this name exists, make sure it is not activated, and then delete or rename the old environment directory (~/.conda/envs/std/).

2. Make sure ~/.condarc looks like this:

   channels:
     - conda-forge
     - defaults
   channel_priority: strict

   This way, packages from conda-forge are accessible and have priority.

3. Create and activate the new environment.

   conda create -n std
   conda activate std

4. Install the Python interpreter and the IPython Kernel.

   conda install python
   conda install ipykernel

5. VS Code tends to remember old environments. To refresh the information, disable the Python extension (and its dependencies, e.g. Pylance), reload VS Code, enable the Python extension et al., and reload. Then let VS Code install what it needs.

6. Install further packages as needed.

Record the state of an environment

The explicitly installed packages (conda install or by VS Code) can be recorded by

conda env export --from-history > std.yml

Recreate an environment from the recorded state

With the file std.yml, the same environment can be recreated somewhere else, follwing steps 1 and 2 and then using

conda env create -f std.yml

Update an environment to a recorded state

If the environment exists, it can also updated using

conda env update --name std --prune --file std.yml