This is a template recipe/feedstock for creating conda packages of GNU Radio out-of-tree modules.

Instructions

Create a new repository using this template and checkout the branch corresponding to the version of GNU Radio that you want to build against. Branch names follow the GNU Radio convention (e.g. "maint-3.8") for building against different major versions (you can ignore the "rendered" branches).

Modify the recipe

First, modify the files in the recipe directory to suit a particular OOT module (the example builds a package for gr-paint).

1. The most important file is the meta.yaml file, where you will have to set the package name and version, update the dependencies, and include informational metadata.
2. You may also have to edit the build scripts, build.sh and bld.bat, to modify the build procedure.

You can then use this recipe to build a conda package using conda build.

Enable CI builds

If you want to create packages through the Github Actions CI infrastructure, there are some additional steps.

1. (optional) Edit the recipe/conda_build_config.yaml file to include your Anaconda.org channel name for uploading packages.
2. (optional) Edit the conda-forge.yml file to include your Anaconda.org channel and Github username.
3. (optional) Set the BINSTAR_TOKEN secret on your Github repository to the value of your Anaconda.org access token. This will enable uploading of the built packages to your Anaconda.org channel.
4. Render the feedstock using conda smithy. From the base repository directory, run the command conda smithy rerender -c auto to create and commit the CI infrastructure files.

To update the recipe in the future, you should create commits on a separate git branch and test them out with a pull request to the base branch. Remember to always re-render when you make changes to the recipe! Once the CI build passes, you can merge the pull request and new packages will be built and uploaded as Github Actions artifacts and (optionally) to Anaconda.org.

Submit to conda-forge

If you create a working recipe, it is highly recommended that you submit it to conda-forge by creating a pull request to the staged-recipes repo adding the recipe. By doing this, you get to take advantage of all of the automation that conda-forge provides, including automatic re-builds for updated dependencies and automatic version updates! It also makes the built packages easier to find and use with the existing gnuradio packages.

Feel free to contact @ryanvolz for help or more information.