Supported Formats

Conda Repository

Cloudsmith provides public & private repositories for Conda

Conda is a cross-platform, language-agnostic binary package manager. It is the package manager used by Anaconda installations.

For more information on Conda, please see:

• Conda Documentation
• Anaconda

Contextual Documentation

The examples in this document are generic. Cloudsmith provides contextual setup instructions within each repository, complete with copy and paste snippets (with your namespace/repo/rsa-key pre-configured).

In the following examples:

Upload a Package

Creating Conda Packages

To upload via the Cloudsmith API/CLI, you'll need to generate a package first. You can do this with conda-build:

conda build .

This generates a .tar.bz2 file like your-package-0.1.0.tar.bz2 that you can upload.

Upload via the Cloudsmith CLI

For full details of how to install and setup the Cloudsmith CLI, see Command Line Interface.

The command to upload Conda package via the Cloudsmith CLI is:

cloudsmith push conda OWNER/REPOSITORY PACKAGE_NAME-PACKAGE_VERSION.tar.bz2

Example:

cloudsmith push conda my-org/my-repo your-package-0.1.0.tar.bz2

Upload via Cloudsmith web app

Please see Upload a Package for details of how to upload via the Cloudsmith web app.

Download / Install a Package

Public Repositories

Adding a new channel can be accomplished by using the conda CLI:

conda config --add channels https://conda.cloudsmith.io/OWNER/REPOSITORY/

Alternatively, the channel can be added directly to your .condarc:

channels:
  - https://conda.cloudsmith.io/OWNER/REPOSITORY/
  - defaults

Or if you prefer, directly in your environment.yml file:

name: env-name
channels:
  - https://conda.cloudsmith.io/OWNER/REPOSITORY/
  - defaults
dependencies:
  - python=3.7
  - codecov

To confirm the channel exists, you can view the channels configured for Conda by using:

conda config --show

Private Repositories

Private Repositories

Private Cloudsmith repositories require authentication. You can choose between two types of authentication, Entitlement Token Authentication or HTTP Basic Authentication. The setup method will differ depending on what authentication type you choose to use.

Warning

Entitlement Tokens, User Credentials and API-Keys should be treated as secrets, and you should ensure that you do not commit them in configurations files along with source code, or expose them in any logs.

Adding a new channel can be accomplished by using the conda CLI:

conda config --add channels https://USERNAME:PASSWORD@conda.cloudsmith.io/OWNER/REPOSITORY/

Alternatively, the channel can be added directly to your .condarc:

channels:
- https://USERNAME:PASSWORD@conda.cloudsmith.io/OWNER/REPOSITORY/
- defaults

Or if you prefer, directly in your environment.yml file:

name: env-name
channels:
- https://USERNAME:PASSWORD@conda.cloudsmith.io/OWNER/REPOSITORY/
- defaults
dependencies:
- python=3.7
- codecov

Install a Package

Once configured, you can install packages from your channel using the conda CLI:

conda install your-package=1.2.3

Or you can search for packages via blended channels (multiple upstreams configured in your Cloudsmith repository) for any package:

conda search your-package

Conda repodata patching

In certain scenarios, Conda package metadata may need to be patched. Patching is often required when a package dependency is updated, and backwards compatibility is broken, as in the following example:

• package-a, version 3.5.0 depends on package-b with a version constraint of package-b >= 2.0.0.
• A new major version of package-b is released (3.0.0). This introduces breaking changes, making it incompatible with all previous versions of package-a.

Repodata patching can be used to resolve the above scenario. For example, the dependency constraint can be amended (patched) to the more specific: package-b >= 2.0.0, <3. This approach does not alter the original package artifacts, and means no requirement is placed on package maintainers to re-release corrected versions of any existing (broken) packages.

Applying repodata patching

Patching is accomplished by applying patching instructions to a channel's repodata file (repodata.json). This file format is used by Conda channels to provide an index of all packages and their respective metadata within a specific subdirectory (subdir) of the channel. When a request is made for the relevant repodata.json file, the patch instructions are merged with the repodata, and the patched (modified) repodata is returned. An overview of this process is available here.

Cloudsmith provides a dedicated cloudsmith-conda-repodata-patches repository to facilitate the easy creation of patching instructions. This repository provides relevant scripts and instructions on how to add, generate and submit your patch instructions to Cloudsmith.

Note

Please note that when repodata patching is applied, either through patch instructions uploaded to the repository or from the upstream, the Cloudsmith repository version takes precedence. Patch instructions are applied on a per-package basis.

Upstream Proxying / Caching

Please see Upstream Proxying for more details.

Upstreams provide a way to blend multiple Conda repositories into a single repository. Cloudsmith supports proxying and caching from upstream repositories for any available channels. This means you can connect not just the canonical upstreams, but also other popular channels. For example, some of the largest upstream channels include:

• Anaconda
• Conda-Forge
• Bioconda

Updated 4 weeks ago