ESP-IDF Component Continuous Integration

[manforowicz]

This pull request is meant to resolve #227 by adding continuous integration for publishing Libcanard to the ESP Component Registry.

Files added:

• idf_component.yml. This file contains required ESP component metadata.
• Kconfig. ESP components use this file as a standard way for controlling build options. I've added options for enabling/disabling assertions and the CRC table.
• CMakeLists.txt. This file is required to register the component with the ESP build system.
• .github/workflows/esp_publish.yml. This workflow uploads the component to the ESP-IDF registry whenever a new GitHub release is published.
• .github/workflows/esp_dry_run.yml. This workflow does a dry-run of uploading to the ESP-IDF registry, without publishing anything. It can only be manually triggered. I've tested this in my repository, and it seems to work.

The GitHub workflows require IDF_COMPONENT_API_TOKEN with an ESP registry token to be added to the repository secrets.

[pavel-kirienko]

Thank you. I have created the token and added it to the repository secrets.

[pavel-kirienko]

Is it possible to hide CMakeLists.txt, Kconfig, and idf_component.yml into a subdirectory, or at least the first two? I don't want to pollute the project root with third-party items.

[pavel-kirienko]

This appears to be doable via repository_info

[manforowicz]

As you suggested, to avoid polluting the root of the repository with ESP-related files, I tried using repository_info and putting the ESP-related files into a subdirectory. However, this didn't work because ESP-IDF gets confused whenever the source code files are not in the directory where CMakeLists.txt is.

So I decided it would be cleanest to just have the continuous integration script copy all the required files into a temporary component directory before publishing. I tested the esp_dry_run.yml workflow locally, and it seems to work correctly.

[pavel-kirienko]

Looks good. Please address the last two nits and we'll merge it.

[pavel-kirienko]

Shouldn't the script go into /esp_metadata?

[pavel-kirienko]

Missing newline at the end

[manforowicz]

Thank you Pavel!