Dear IANVS users,

we are deprecating the Anaconda module and replacing it with Micromamba, a small, fast alternative implementation of Anaconda’s package manager, Conda. We have also added a short tutorial on how to use Micromamba environments.

Why?

Anaconda (née „Anaconda Python Distribution“) is a repository of packages used for scientific computing, ready-made for installation with Anaconda’s package manager, Conda. Conda provides a feature called environments, which allow projects to have different dependencies without interfering with each other. However, the default installation variant of Anaconda, which we used, comes with a base environment, which contains Conda itself as well as a preinstalled collection of packages. This base environment is the same for all users.

We expected users to ignore the base environment because, in our eyes, per-project environments have advantages that make them clearly superior. However, we failed to communicate this expectation, and inevitably, some users decided to work in the base environment instead of creating their own. At that point, any updates to the base environment—including updates to the Conda package manager itself—could break these users‘ workflows and make their experiments non-reproducible, which is something that we try very hard to avoid.

Micromamba solves this issue by simply not providing a base environment. It is implemented in C++, which makes it fast, and consists of a single statically linked file, which makes updating it trivial. Usage is straightforward if you are familiar with Conda:

# Load the module
$ ml micromamba

# Create a new environment that uses Python 3.9
$ micromamba create -n my-project python=3.9
[...]

# Activate the environment
$ micromamba activate my-project

# Check Python version to confirm
(my-project) $ python --version
Python 3.9.18

In fact, the usage is so similar that we have decided to provide ‚conda‘ as an alias to ‚micromamba‘:

# Load the module
$ ml conda

# Create a new environment that uses Python 3.9
$ conda create -n my-project python=3.9
[...]

# Activate the environment
$ conda activate my-project

# Check Python version to confirm
(my-project) $ python --version
Python 3.9.18

For more information see the aforementioned tutorial.

What happens to the Anaconda module?

Nothing. As mentioned before, we take backwards compatibility and reproducibility seriously. However, the old Anaconda module will be marked as deprecated, which means that it will emit a message whenever it is loaded, pointing users toward the micromamba module instead.

As always, please contact us if you have any questions: via e-mail or phone, (0345) 55-21864 or (0345) 55-21868.

Best regards,

ITZ HPC Team

ITZ | Keine Kommentare

Schlagworte: HPC | IANVS

Über Patrice Peterson

• Mitarbeiter/in
• Website: http://itz.uni-halle.de/hpc