Skip to content

Conda and Miniconda

CloudRepo hosts Python packages rather than conda packages or a conda channel. A conda user installs them with pip inside the conda environment. This page shows how to do that without letting a package from PyPI replace one of yours.

  • A Python repository to install from and publish to. Creating a Repository shows how to make one.
  • A repository token that reaches it, and Download and publish packages access if you will publish. Repository Tokens: Create One and Authenticate shows how to mint one. pip sends the email address of the account that minted it as the username, and the token as the password.
  • conda or Miniconda, installed from conda’s own instructions.

Install private packages into an environment

Section titled “Install private packages into an environment”

Make the environment with pip in it, activate it, and install your private packages from CloudRepo alone, in a pip step of their own. Anyone can publish a package on PyPI under your private package’s name, and a pip step that can also read PyPI may install theirs instead.

Terminal
conda create --name myproject python=3.12 pip
conda activate myproject

Write requirements-private.txt as in Install private and public packages together: its first line names the CloudRepo index, and pip finds the login for the host in your ~/.netrc, which that page also shows how to write, so the file stores no secret and can be committed. Then, with the environment active:

Terminal
pip install -r requirements-private.txt

Everything that applies to pip applies here, because the install is pip’s. Public packages go in a step of their own (pip install -r requirements.txt, from PyPI), nothing in that step may depend on a private package, and the public packages your private packages depend on go in requirements.txt too: the CloudRepo step does not look for them on PyPI, and fails if one is missing. The full rules, the three-step order and what a hash file does are on Python Repositories. If an earlier setup put CloudRepo and PyPI together in pip.conf, pip config or PIP_EXTRA_INDEX_URL, remove that: every pip install reads such a pair as one index.

Combine conda packages and private packages in environment.yml

Section titled “Combine conda packages and private packages in environment.yml”

conda installs the pip: section of environment.yml with a single pip install that covers every line in it. So keep that section for public packages, and install your private packages in the second step above, after the environment exists. Do not put the CloudRepo URL or a token in environment.yml:

environment.yml
name: data-science-project
channels:
- conda-forge
dependencies:
- python=3.12
- numpy
- pip
- pip:
- requests
Terminal
conda env create --file environment.yml
conda activate data-science-project
pip install -r requirements-private.txt

Publishing is pip’s side again: install twine in the environment and upload with it, as in Publish Python packages to CloudRepo. A wheel built inside a conda environment is an ordinary wheel, and CloudRepo stores it like any other Python file.

The step that reaches CloudRepo is the pip install -r requirements-private.txt above, so that is where a refused token shows. pip does not print a 401 for a token it read from ~/.netrc. It asks who you are instead:

User for your-org.mycloudrepo.io:

A revoked or expired token does this, and so does a machine or login line that does not match. With no one at the terminal to answer, as in CI, the install ends there with EOFError: EOF when reading a line.

The other messages are pip’s, and they mean what they mean outside conda: see Install Python packages from CloudRepo. If the CloudRepo step stops because a public package is missing, install that package in the public step first: the CloudRepo step does not look for it on PyPI.

See also Repository tokens, which covers the credential each client uses, and Python Repositories, which covers pip, twine and the Python repository settings.

View as Markdown