conda environments

In the course we will work with a conda environment. An environment contains a specific collection of conda packages that you have installed.

You should always work within a dedicated environment, rather than the default (base) environment. This allows to separate the packages used for analyses and for handling packages etc. (e.g. conda itself is installed in the base environment).

You should have successfully finished the instruction in Using conda before continiuing here. We will again work on a command-line terminal:

Windows users should open the Miniforge prompt (search in the start menu)
macOS and linux users open a shell (Konsole or terminal)

(Note: users of JupyterHub can start a terminal under File ‣ New ‣ Terminal)

At this point you should have a window showing the following:

(base) username@computer:~$

Analysis environment

For the Integrated Practical - Python in Geosciences course we will work in a environment with the name ipp_analysis. We create this environment from a file that lists all required packages.

In the section Preparation we created a folder for the course named ip_python.
From the terminal we now need to navigate to this folder. This can be done using the cd command:
```
cd ip_python
```
Depending on where you created the folder you need to adapt the path, e.g.:
```
cd Documents/ETH/ip_python
```
Next we need to download the environment.yml file.
- Download the file and save it to the ip_python folder.
  
  Warning
  
  Sometimes the file can be renamed to environment.yml.txt by the download (especially on macOS). Rename it back to environment.yml.
Open the environment.yml file in a text editor and take a look. Which packages will be installed?

Setup the analysis environment

Next we create the environment that we will use for the course. This is also done from the terminal:

conda env create --file environment.yml

Tip

You need to be in the same folder as the environment.yml file else it does not work.

This will download and install the packages listed in the environment file. It contains all packages used for the introduction and the projects (except the climada project). This is still part of the preparation for the python exercises. If you are having trouble please come ask us for support.

Test environment

It might still be unclear why we are using environments to install python and python packages. In this section we will try to motivate this and also get a bit more familiar with conda environments. For this we set up a small test environment. It’s not really needed for the course so we will delete it again at the end.

All commands have to be entered in the Terminal opened in the last section.

Above we created a environment from a file that listed all packages we wanted. Here, we specific the packages manually. We do this using the following command to install python and numpy:
```
conda create --name test_env python=3.10 numpy=1.21
```
This creates a new environment with the name test_env. We specify two packages that should be installed, namely python, and numpy. It will install more than these two packages, because they have dependencies that will also be added. By using python=3.10 and numpy=1.21 we tell conda to install a specific version of python. If no version constraint is given the newest compatible version is installed.
Now that the environment was created we can activate it:
```
conda activate test_env
```
Now your command prompt should show that your test_env environment is active:
```
(test_env) username@computer:~$
```
Once it is active we can install additional packages into the environment. For example we can add the python plotting library matplotlib:
```
conda install matplotlib
```
Again, this should install matplotlib as well as additional dependencies.
To see all the packages that are currently installed, their version and source channel you can execute
```
conda list
```
from the command prompt.
We can also list a specific package:
```
conda list numpy
```
This should indicate that numpy version 1.21 is installed. If we compare this with the numpy version installed in the ipp_analysis environment:
```
conda list -n ipp_analysis numpy
```
we should see that it contains a different version. Being able to have different versions of python and packages installed, and to be able to switch efortlessly between is one of the main motivations to work with environments.
We leave the test_env environment again by calling:
```
conda deactivate
```
You show all environments that you installed using:
```
conda env list
```
Environments can be removed by calling
```
conda env remove --name test_env
```
You can remove the test environment - it’s no longer needed.