6 Commits

Include renames

Author	SHA1	Message	Date
Nick	9731181a6b	fix(tests): respawn the bridge daemon on mid-test death (root-caused) ... The hardening's own diagnostic nailed it (run 18359): the daemon answers /health at fixture setup, then dies SILENTLY mid-test (alive=False, out='') on the saturated runner — host shows 20GB swap in use, and a silent kill with zero output is the memory-pressure signature, not a race. A retry can never survive a dead process. Fix: registrations are file-backed under WR_ROOT, so a respawned daemon resumes the same state — post_event now ensures liveness before each attempt (respawn on death) and _spawn is the single spawn path (init's T21.5 fresh-port retry preserved). Bridge suites 19/19, ruff clean.	2026-10-04 12:30:37 +01:00
Nick	544c19719e	fix(tests): harden the bridge daemon post_event against load-dependent refusal ... Observed 2026-10-04 (runs 18340/18342/18346/18347): the fixture's daemon answers /health at setup, then a POST lands ECONNREFUSED under runner load — the twin-failure pattern across two commits while sibling runs passed. Harness-side hardening per the intermittent-defect law: post_event retries twice with a beat, and on final failure dumps the daemon's captured output (alive state + log tail) so the next diagnosis starts from evidence. Bridge suite 11/11 locally, ruff clean.	2026-10-04 12:18:28 +01:00
Nick	756ebe6201	fix(t21.5): pyright-clean provider loader; harden test Daemon fixture against bind races (retry on fresh ephemeral port)	2026-09-17 13:06:17 +01:00
Nick	e1a71efeb9	feat(ci): event-driven CI qualification wait over the existing W/R bridge (T16 16C/16D/16E) ... - deploy.yml sleep-30 polling loop REPLACED: registers a gitea-identity PROCESS_EVENT wait, yields in select() on the FIFO, resumes on the ci.yml notify-event-bridge WAKE, then re-reads the canonical Gitea verdict (deploy_qualify.py). Event payload is never completion truth. - ci.yml gains notify-event-bridge (always(), best-effort): a red CI wakes the waiter as loudly as a green one; bridge unreachable -> watchdog recovery covers the lost wake. - event_wait.py: register_process_wait gains provider/event_type/sha identity (16E: ONE bridge primitive; builder + gitea are adapters); open_fifo_reader + caller-owned fifo_fd close the 16Y#10 race (event before waiter in select()); await_resume_event returns eof instead of busy-looping on a caller-owned fd. - ci_event_wait.py watchdog: failure DETECTOR only — liveness-gated re-arm, one-shot authority recovery on lost events, truthful failure when no verdict exists; timestamps emitted for the 16D latency proof. - tests: 8 held-outs on the REAL daemon (shared wr_daemon fixture in conftest — no duplicated helpers). Daemon-side: /process-event passes through provider/event_type/sha (defaults preserve T15); lease_path sanitizes slash-bearing session ids (resume delivery was impossible for them — found live, fixed in daemon snapshot 79434e0). Suite 1861 green; ratchets PASS; Henxels rc=0.	2026-09-15 16:26:54 +01:00
Nick	fd5f31bba2	fix(scaffolding): §15D tests use ephemeral ports, not fixed 8793 ... CI failure on the 0.66.61 bump run (job 18584): the latency test got watchdog/alive:False because its daemon could not bind the FIXED port 8793 — a stale TIME_WAIT socket from an earlier test in the same file held it (verified locally: 8793 bindable only with SO_REUSEADDR while no listener exists). The notifier's event POSTs went nowhere, so the watchdog correctly reported PROCESS_LOST: the harness was at fault, not the bridge. Daemon fixture now binds port 0 (ephemeral) and reads back the assigned port. 3x consecutive green locally.	2026-09-15 14:14:44 +01:00
Nick	eccb446a3c	feat(scaffolding): §15C/§15D/§15E — event-driven wait becomes the canonical default ... - 15C audit: the W/R bridge (daemon :8791) was complete but its resume target was mimo-session-shaped only — every Python controller had to sleep/poll. Root cause recorded, not excused. - src/rune/event_wait.py: register_process_wait (durable PROCESS_EVENT wait + per-wait FIFO resume cmd), run_builder_under_events (real builder under a notifier that observes actual exit and POSTs the event), await_resume_event (select-block, NO sleep floor, watchdog deadline = failure detector), resume_or_recover (restart: durable waitstore wins over transient token). - 15E adversarials: duplicate/stale/wrong-correlation/malformed events, controller restart, lost-event watchdog (zombie-aware liveness). - Latency proof: builder exits at T, controller resumes at T+~0.1s — no arbitrary floor. Failures carry truth (exit_code verdict). - tests/test_event_wait.py: 8 proofs, real daemon subprocess, disposable WR_ROOT. Daemon invariant suites 45/45 + 36/36 unchanged.	2026-09-15 02:12:54 +01:00