feedstock-outputs

registry of feedstock outputs

Configuration

The JSON blob config.json contains the configuration for the feedstock outputs registry. The configuration contains the following fields:

• outputs_path: the path to the sharded outputs in the repo
• shard_level: how deeply the outputs are sharded
• shard_fill: the character to use to fill in sharding levels for outputs with names shorter than shard_level
• auto_register_all: whether to automatically register new outputs for any feedstock

The auto_register_all boolean is used by the conda-forge-webservices and conda-forge-ci-setup packages to determine if new outputs should be automatically registered. If auto_register_all is true, then any new outputs for feedstock will be automatically added. If it is false, then only those that match the glob patterns (as described below) will be added.

Format

Each json blob maps the conda package (in the name of the json blob) to the feedstocks that produce it (in the "feedstocks" list in the json blob contants).

Automatic Registration of Outputs that Follow Regular Patterns

Some feedstocks output packages that have a regular pattern (e.g., llvdmdev outputs libllvm19, libllvm20, etc.). We store a list of allowed output glob patterns in this file. Each entry is a list of allowed glob patterns for the feedstock. These patterns are tested against new outputs using the Python fnmatch module. If a feedstock uploads a package that matches one of these patterns, the output is automatically registered.

Adding New Outputs and Glob Patterns

If you make a change so that a package is produced by a different feedstock or adds an output, please make a PR using the admin-requests repo to add the new output. You can add new glob patterns via the admin-requests repo as well.