Say that a free ADR number means free across open PRs too #45

Merged
rob merged 2 commits from docs/adr-numbering-across-open-prs into main 2026-08-04 21:39:54 +00:00
Owner

Two ADR number collisions today (0039, then 0043). In both cases each author correctly took "the next free number" from main — and got the same answer, because neither branch had merged.

Records that "next free" spans main plus every open PR, and that renumbering is forbidden only after a record lands on main, where something might link to it. Before merge, renumbering the branch further from merging is the fix.

Documentation only.

Two ADR number collisions today (0039, then 0043). In both cases each author correctly took "the next free number" from `main` — and got the same answer, because neither branch had merged. Records that "next free" spans `main` plus every open PR, and that renumbering is forbidden only *after* a record lands on `main`, where something might link to it. Before merge, renumbering the branch further from merging is the fix. Documentation only.
Say that a free ADR number means free across open PRs too
All checks were successful
CI / build (pull_request) Successful in 2m8s
267e609d55
rob left a comment

Verdict: mergeable

No contradiction with ADR-0001 — it already scopes "never renumber" to a record merged onto main ("This rule binds from the moment a record is merged onto main"), so this change is a correct restatement, and actually tightens the README's previously unqualified "never renumber them" to match. Worth citing ADR-0001 here for that reason, the way the partial-supersession paragraph just below already does — right now the two texts state the same rule with no link between them.

History check: the 0039 collision is well corroborated — ADR-0039 (Editor-delete-places) merged at 17:57 on 2026-08-04, and "Renumber password policy ADR to 0040" landed 16 minutes later, i.e. the not-yet-merged branch was the one renumbered, matching the guidance exactly. I could not find any commit evidence (message, or a second "Renumber…" commit) for the claimed 0043 collision — that doesn't mean it didn't happen (a collision caught before push leaves no separate trace), but it's the one claim in the PR body I can't independently confirm. Worth double-checking before merge since the whole point of the change is to be a reliable factual record.

Guidance is actionable as written ("check the open PRs... say in the PR body which number you took").

Verdict: mergeable No contradiction with ADR-0001 — it already scopes "never renumber" to a record merged onto `main` ("This rule binds from the moment a record is merged onto `main`"), so this change is a correct restatement, and actually tightens the README's previously unqualified "never renumber them" to match. Worth citing ADR-0001 here for that reason, the way the partial-supersession paragraph just below already does — right now the two texts state the same rule with no link between them. History check: the 0039 collision is well corroborated — ADR-0039 (Editor-delete-places) merged at 17:57 on 2026-08-04, and "Renumber password policy ADR to 0040" landed 16 minutes later, i.e. the not-yet-merged branch was the one renumbered, matching the guidance exactly. I could not find any commit evidence (message, or a second "Renumber…" commit) for the claimed 0043 collision — that doesn't mean it didn't happen (a collision caught before push leaves no separate trace), but it's the one claim in the PR body I can't independently confirm. Worth double-checking before merge since the whole point of the change is to be a reliable factual record. Guidance is actionable as written ("check the open PRs... say in the PR body which number you took").
Cite ADR-0001 for the immutability the renumber rule mirrors
All checks were successful
CI / build (pull_request) Successful in 2m12s
77d93a797e
rob merged commit 3633ad1eeb into main 2026-08-04 21:39:54 +00:00
rob deleted branch docs/adr-numbering-across-open-prs 2026-08-04 21:39:55 +00:00
Sign in to join this conversation.
No reviewers
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
rob/PlaceMark!45
No description provided.