dependabot: apply cooldown except for action reviews#712
Conversation
Try claude.... (and not joking about it) - for things like renaming workflows etc. It's actually very good in findng all the places that needs to get updated :) |
a8c0087 to
2f71107
Compare
|
Fixed a few things, now seems OK to me. That "Verify Dependabot Action Build" is skipped makes sense since this is not a dependabot PR |
potiuk
left a comment
There was a problem hiding this comment.
Following up on my earlier review — turning the actionable items into inline suggestions. Three of these (the using: "composite" header, the update_composite_action.yml typo, and the open-pull-requests-limit: 50 placement) are blockers in my opinion; the rest are polish.
bed8e23 to
29e75af
Compare
29e75af to
e488982
Compare
potiuk
left a comment
There was a problem hiding this comment.
Re-review on e488982 — all three blockers and every polish item from the previous pass are addressed. Confirmed against the new HEAD:
Blockers
- ✅
gateway/gateway.py:183—using: "composite"is now in the header literal. - ✅
update_composite_action.yml— the singular/plural mismatch is resolved by renaming the file (the opposite direction from my suggestion, which works just as well since the in-file references are now consistent). - ✅
.github/dependabot.yml—open-pull-requests-limit: 50moved onto the allowlist-review block.
Polish
- ✅
update_refs:dummy_steps→composite_steps(signature + docstring + body). - ✅
update_actions: docstringPath to the dummy workflow file→Path to the composite action file. - ✅
update_workflow:dummy_path→composite_action_path, docstring updated, andwrite_strcall fixed to match. - ✅ Bonus:
name: Update Dummy workflow→name: Update Approved Patterns and Composite Action.
Still open (non-blocking, from the prose review)
cpp-linter/cpp-linter-action >=2.16ignore:is only on the allowlist-review block. In practice it's fine today (the action isn'tuses:-d anywhere in the real workflow tree), but if that ever changes the ignore won't apply — worth a durable comment or a duplicatedignore:on the second block.- The
leafo/gh-actions-luarocksre-ordering inactions.ymlis unrelated to the cooldown refactor. Consider splitting it out for a cleaner history, or at least calling it out in the PR description. - Optional: a one-line snapshot test asserting
'using: "composite"' in generate_composite_action({})so the regression caught here cannot sneak back in.
LGTM apart from those nits.
Fixes #683
Sadly GHA are not so easy to test, so this might take some rounds to get right :)