Github actions CI #515
Github actions CI #515
Conversation
Avoids a warning, which is treated as an error on some Sphinx versions.
This changes the CI to be completely cloud hosted and independent of the runner machine. I think all features are retained and the .docker directory can be blasted away. Note that soon the buildmachine will stop building F* images, so the old workflow will go through the 'standalone' path and be slower.
Who is in charge of regenerating the base container? Is this something official that the F* team commits to maintaining? Should we switch other projects like HACL* to it? |
Currently I'm not regenerating it at all (it's described in I think the right thing to do is just set the environment up with specific actions instead of rolling our own container, and write a |
This changes the CI to be completely cloud hosted and not rely on state in the runner machine. I think all features are retained and the .docker directory can be blasted away. Except for Slack messages, I didn't port that chunk as it will not work over multiple jobs, and it's probably better if we find something canned instead of retaining the old code. (Plus maybe move to Zulip?)
It does use a base container hosted in Dockerhub, at least for now. It'd be nicer to keep a container description in the repo and build on-demand.
Note that soon the buildmachine will stop building F* images, so the old workflow will start going through the 'standalone' path and be slower.
Example run: https://github.com/mtzguido/karamel/actions/runs/12663210092