The project's dependencies are encoded in the Conda environment file environment.yml, allowing developers to easily set up a local environment for testing/developing HTSinfer.
On the other hand, when building a Docker image from the provided Dockerfile, dependencies are installed via apt and pip, not Conda.
This leads to discrepancies between environments being set up locally, during CI and during use. Moreover, dependency versions need to be maintained in two different locations.
Solution
Use Conda to build the environment inside the Docker image.
Problem
The project's dependencies are encoded in the Conda environment file
environment.yml
, allowing developers to easily set up a local environment for testing/developing HTSinfer.On the other hand, when building a Docker image from the provided
Dockerfile
, dependencies are installed viaapt
andpip
, not Conda.This leads to discrepancies between environments being set up locally, during CI and during use. Moreover, dependency versions need to be maintained in two different locations.
Solution
Use Conda to build the environment inside the Docker image.