rdmsm4x - changelog - ReplicantDB hang fix, Settings+Status shell, P0 index fix, Builds 45-52 fleet-wide - claude - ISSUE-20260926-36 - replicantDB - 20260927-0243
Summary: ReplicantDB's "Run Maintenance Pass Now" no longer beachballs (measured on the installed app). The app gained a Settings window, a Status-first shell, color profiles and glass, badges and notifications, a provider catalog and a permissions panel. A P0 was fixed: no new files had been indexed since 09-01. Builds 45, 47 and 49–52 shipped Developer ID-signed and notarized to all six Macs.
Scope
- Hosts: rdmsm4x, rdmpw3275m, rdmpw3265m (Intel 26.7.1), rdmbair13m5, rdmbair15m5, jdmbair13m5.
- Repo: rdmsm4x:~/dev/apps/replicantDB. main
ff76a00; tags v1.21.0-build47, v1.21.1-build49, v1.22.0-build50, v1.22.1-build51, v1.22.2-build52; pushed to fleet and backup (private). - Session: claude@rdmsm4x/rdbhang1, 2026-09-26 23:59:53 → 2026-09-27 02:43 EDT.
What changed (on every host)
- /Applications/ReplicantDB.app, ~/bin/replicantdb-cli and ~/bin/replicantdb-mcp replaced with 1.22.2 (52), exe sha256 47819aba…, via transactional rollout. Prior GUI and daemon states were preserved.
- A rollback copy per run is under ~/Library/Application
Support/ReplicantDB Rollbacks/
/ on each host. - Index DB: FTS triggers rewritten to the on-disk columns. 1,833 ~/dev/arista rows purged per DEC-20260927-03. Additive table extraction_failures, created only when a failure is recorded.
Verification
- Suite counts per build: 45 1,475/1,476 (1 timing assertion at load 264, passes alone), 47 1,512/0, 49 1,536/0, 50 1,556/0, 51 1,556/0, 52 1,559/0.
- Notarization accepted for each build; stapled; Gatekeeper "Notarized Developer ID"; designated requirement identical to Build 44.
- The hang, on installed Build 45: click 0.27 s; max AX round-trip 1.27 s across a pass including its completion (Build 44: 1.67/2.29 s plus a synchronous ~7 s getStats on main).
- P0 on rdmpw3275m: 0 new rows 09-02→09-27; 13,482 new rows within 17 min of Build 47.
Undo
- Per host: restore from ~/Library/Application Support/ReplicantDB
Rollbacks/
/old (rollout script rollbackaction). - Source: git revert on main.
Owner actions
- Rich: purge the ~1.07 M existing ~/Library index rows? (Mail 509,500 / CloudStorage 527,225).
- CloudKit container registration is queued (possible keychain prompt).
- Notes publish of this entry is pending: queued so no Automation prompt fires overnight.
Evidence
~/dev/_handoff/replicantdb-hang-20260927/ (HANDBACK.md, BEST-PRACTICES-REVIEW.md, shots/, release/).