I have a project where I have to mix conda dependencies that are publicly available with dependencies from my own private pip package registry. These are all pip dependencies. It is possible to define a pip dependency like this in the env.yml file:
name: env-name
channels:
- anaconda
- conda-forge
- defaults
dependencies:
- conda-dependency=1.0.0=s0mel3tters
- pip:
- --extra-index-url https://__token_name__:__token__@gitlab.com/path/to/registry
- pippackage==1.0.0
This works. Great. But. I don't want to check my __token_name__
and __token__
in source control for obvious reasons. Now it turns out that I can define environment variables into the env.yml that will be created in the actual environment. I want to do something slightly different. I want to read in an environmental variable that already exists and use it in my pip URL definition. Is this possible? If so, how?
Apparently, you can substitute for environmental variables:
name: env-name
channels:
- anaconda
- conda-forge
- defaults
dependencies:
- conda-dependency=1.0.0=s0mel3tters
- pip:
- --extra-index-url https://${TOKEN_NAME}:${TOKEN}@gitlab.com/path/to/registry
- pippackage==1.0.0