.hyperi-ci.yaml:72 still keys the release block as publish: where hyperi-ci now spells it release:. Nothing breaks -- the old spelling is accepted and prints a deprecation warning -- so this is whenever-it-suits work.
Filing it because culvert is the last repo in the suite with no issue for it, and the gap is the point rather than the config line.
The same migration is tracked everywhere else it applies: dfe-receiver#122, dfe-transform-vrl#64, dfe-transform-vector#80, and dfe-engine#424 which is now closed because that repo already migrated. scalo-py landed its own as #34. So the convention is held in some repos, tracked-but-unheld in three, and until now silently unheld here.
That shape is the actual defect. A convention that is true in most places and quietly false in one is harder to spot than one that is false everywhere, because every sample you check agrees with you. It is worth fixing here mainly so the fleet sweep has no silent member.
The change, from origin/main:
publish: -> release:
enabled: true enabled: true
channel: release channel: release
Done when .hyperi-ci.yaml keys the block release: and a run emits no deprecation warning for it.
Found while checking the suite for repos carrying the legacy key, after hyperi-ci's own sweep touched the same file in six sibling repos.
.hyperi-ci.yaml:72still keys the release block aspublish:where hyperi-ci now spells itrelease:. Nothing breaks -- the old spelling is accepted and prints a deprecation warning -- so this is whenever-it-suits work.Filing it because culvert is the last repo in the suite with no issue for it, and the gap is the point rather than the config line.
The same migration is tracked everywhere else it applies: dfe-receiver#122, dfe-transform-vrl#64, dfe-transform-vector#80, and dfe-engine#424 which is now closed because that repo already migrated. scalo-py landed its own as #34. So the convention is held in some repos, tracked-but-unheld in three, and until now silently unheld here.
That shape is the actual defect. A convention that is true in most places and quietly false in one is harder to spot than one that is false everywhere, because every sample you check agrees with you. It is worth fixing here mainly so the fleet sweep has no silent member.
The change, from
origin/main:Done when
.hyperi-ci.yamlkeys the blockrelease:and a run emits no deprecation warning for it.Found while checking the suite for repos carrying the legacy key, after hyperi-ci's own sweep touched the same file in six sibling repos.