Purpose
Tracking home for ADR-0013 (decisions/0013-repository-settings-baseline.md), covering:
- The maintainer disposition record for its acceptance (below), since
runbooks/adr.md requires that acceptance record a named decider and that the ADR's own PR/issue trail make the authorization visible -- unlike ADRs 0008/0009/0010/0011, ADR-0013 had no pre-existing GitHub-native decision-packet comment; the disposition below fills that gap.
- Rollout tracking per the ADR's own "Rollout and rollback" section: a read-only audit of current branch protection/ruleset settings across all 86 active repositories (step 3), followed by deliberate, gated, per-repository application (step 4). No repository setting is changed by the ADR itself or by this issue on its own.
Maintainer disposition (2026-07-25)
Decider ss-o. PR #474 proposed ADR-0013 with one open question: whether class-1 (continuously-deployed) repositories should require linear history on their deployed branch.
Decision: drop the requirement for class 1. Issue #473 (merged the same day PR #474 was opened) documents that class-1 repositories promoting a persistent development branch (ADR-0008's next) into the deployed branch diverge after every promotion when linear history is required -- squash and rebase both mint commits the development branch does not have, and a merge commit, the one strategy that preserves ancestry, is exactly what linear history forbids. The only available fix was a direct, admin-bypassing push, and #473 records a near-miss where the bypass warning was silently truncated by piping through tail. Classes 2-4 are unaffected: their promotions are either infrequent (class 2 release tags) or nonexistent (trunk-only classes 3-4).
ADR-0013 is ACCEPTED on this basis, same authority and process as ADRs 0008/0009/0011 decided the same day. No repository configuration changes are authorized by this decision -- rollout stays gated per the ADR's own step 1 ("Accept this ADR. It changes no repository by itself").
Next: read-only settings audit
Per rollout step 3, the next step is a read-only survey of branch protection and ruleset settings across all 86 active, public, non-fork repositories, reusing PR #474's own audit methodology so the numbers reconcile with its 75/86 finding. Findings will be posted to this issue. No repository setting will be changed by the audit itself.
Purpose
Tracking home for ADR-0013 (
decisions/0013-repository-settings-baseline.md), covering:runbooks/adr.mdrequires that acceptance record a named decider and that the ADR's own PR/issue trail make the authorization visible -- unlike ADRs 0008/0009/0010/0011, ADR-0013 had no pre-existing GitHub-native decision-packet comment; the disposition below fills that gap.Maintainer disposition (2026-07-25)
Decider ss-o. PR #474 proposed ADR-0013 with one open question: whether class-1 (continuously-deployed) repositories should require linear history on their deployed branch.
Decision: drop the requirement for class 1. Issue #473 (merged the same day PR #474 was opened) documents that class-1 repositories promoting a persistent development branch (ADR-0008's
next) into the deployed branch diverge after every promotion when linear history is required -- squash and rebase both mint commits the development branch does not have, and a merge commit, the one strategy that preserves ancestry, is exactly what linear history forbids. The only available fix was a direct, admin-bypassing push, and #473 records a near-miss where the bypass warning was silently truncated by piping throughtail. Classes 2-4 are unaffected: their promotions are either infrequent (class 2 release tags) or nonexistent (trunk-only classes 3-4).ADR-0013 is ACCEPTED on this basis, same authority and process as ADRs 0008/0009/0011 decided the same day. No repository configuration changes are authorized by this decision -- rollout stays gated per the ADR's own step 1 ("Accept this ADR. It changes no repository by itself").
Next: read-only settings audit
Per rollout step 3, the next step is a read-only survey of branch protection and ruleset settings across all 86 active, public, non-fork repositories, reusing PR #474's own audit methodology so the numbers reconcile with its 75/86 finding. Findings will be posted to this issue. No repository setting will be changed by the audit itself.