Repository navigation
Add a workflow to regularly run make linkcheck #917
Description
This is a follow-up of #727 and related to #681. Since make linkcheck is somewhat unreliable and can't be required for regular PR, the idea is to use a workflow to run it regularly and create (or update) an issue when a link is unreachable. Note that make linkcheck is already run for PRs, but the result is only visible by looking at the logs.
@AlexWaygood shared a similar workflow in #727 (comment):
Over at typeshed, we have a test that runs every night. If it fails, a bot auto-creates an issue for us. It works pretty well!
Here's the .yml file: https://github.com/python/typeshed/blob/master/.github/workflows/daily.yml
And here's an example of an issue that the bot auto-created when the test failed: python/typeshed#7572
This workflow creates a new issue every time, and it might be ok if it's run weekly/monthly however -- if we want to run it more frequently -- updating an existing issue might be a better approach.
@hugovk also raised a concern in #727 (comment)
One concern: if the cron is often failing (which it might, and often), we don't want it to show the repo as failed at the top:
If this sounds useful we can agree on the details, fix the currently broken links, and adapt the workflow linked above.
Activity
CAM-Gerlach commentedon Jul 30, 2022
FWIW, for spyder-ide/spyder-docs which is of the same OOM as this repo, we do run it for every PR, and periodically (every couple months) update any redirects. However, I certainly wouldn't run it for a repo much bigger than this, like the PEPs or docs, and I could certainly see why that wouldn't be desirable even at this scale, depending on the number and fickleness of the links involved, as well as the obvious workflow considerations.
I believe you could make the action itself not mark itself as failed if the linkcheck fails, but just open the issue, so the whole repo doesn't go red?