The tidy-check failure is unrelated - but this failure does remind me that we should be running clang-tidy on all files (not just on the changeset of PRs) in the nightly tests, as failures can sneak in and remain undetected otherwise. CC @carlopi

The tidy-check failure is unrelated - but this failure does remind me that we should be running clang-tidy on all files (not just on the changeset of PRs) in the nightly tests, as failures can sneak in and remain undetected otherwise. CC @carlopi

I think this is already the behaviour (sort of), but this do not accounts for feature (and there are no nightly run on feature on any case).

Example: on last night CodeQuality run (run on main) the complete version Tidy Check step (example: https://github.com/duckdb/duckdb/actions/runs/11356627641/job/31588215423, , 1h 16') has been taken, while on regular PR the Tidy Check Diff step is taken (example: https://github.com/duckdb/duckdb/actions/runs/11349485053/job/31565802783, 55 seconds).

I will add the complete step also on feature, that is easy but has no immediate effect.
We can talk about how / when to trigger CI also on feature. But at least that should allow to trigger it on request.