--- status: historical created: 2026-08-26 archived: 2026-08-26 reason: "Measurement record: the default-app reload STEP's post-#6292 ON gate ran 9/10 green; the only red was a new store-incomplete setup-error shape, so the skip stayed." --- # Default-app reload post-#6292 re-measure This report records the ten-run re-measure of the oldest remaining patterns STEP skip after #6292 fixed the client-side loss of `cid:` effects before the initial watch response. The gate targeted `integration/default-app.test.ts`'s `should persist and reload every rapidly created notebook note` step. - Repository: `commontoolsinc/labs` - Worktree: `/Users/berni/labs-worktrees/r01-remeasure` - Branch: `claude/server-exec-v2-r01-remeasure` - Measured head: `37b45336a6b17ad27039cc525e4ba2e89f517449` - #6292 commit in the measured head: `9c9073995` - Pinned toolchain: `/Users/berni/.local/share/loom/toolchains/deno/2.9.4/deno`, Deno 2.9.4 - Evidence root: `/Users/berni/labs-worktrees/r01-remeasure-evidence` ## Method The campaign followed the measurement method in `ow45-armb-server-ensure-stage1-report.md` exactly: - Built one ON toolshed binary at the measured head. - Neutralized only the target step's skip guard in the working tree for all runs; the registry entry remained present. - Verified the target step ran by parsing its exact verdict after the step description; no skip-print appeared in any counted run. - Used a fresh store, an independent 97xx port, and a posture probe per run. - Required `shellServerExecutionDefine == "true"` and a present `servingLoop` before starting the test. - Set `SERVER_EXECUTION_ENSURE_SPACE_ROOTS=false`, matching the continuous integration ON lanes where the entry runs. - Masked the LLM environment with `CFTS_AI_GATEWAY_URL=""` and `CFTS_AI_LLM_ANTHROPIC_API_KEY=fake`. - Ran from `packages/patterns` with the prescribed unchanged 600-second harness bound. - Interleaved five quiet and five loaded counted runs and recorded the machine's actual load averages. - Tore down by PID only and verified the port was free after every run. ## Build - Command: `COMMIT_SHA=37b45336a6b17ad27039cc525e4ba2e89f517449 EXPERIMENTAL_SERVER_EXECUTION=true deno task --no-lock build-binaries toolshed` - Binary: `/Users/berni/labs-worktrees/r01-remeasure/dist/toolshed` - Size: 619,556,642 bytes - SHA-256: `747c162b30bd18e144ebbf9ef1c03b7a84d44d005949bcbd4a919e17d1970ebd` ## Counted-run ledger | Run | Mode | Port | Load before / after | rc | Wall | Step verdict | Catch-up | Catch-up failed | Pattern-load error | Stale-read lines | Evidence | | --- | --- | ---: | --- | ---: | ---: | --- | ---: | ---: | ---: | ---: | --- | | r01 | quiet | 9741 | 16.44 / 20.51 | 124 | 600 s | TIMEOUT | 1 | 0 | 0 | 1 | `/Users/berni/labs-worktrees/r01-remeasure-evidence/runs/r01/` | | r02 | loaded | 9742 | 14.07 / 32.26 | 0 | 41 s | GREEN (`ok`, 24 s) | 2 | 0 | 0 | 2 | `/Users/berni/labs-worktrees/r01-remeasure-evidence/runs/r02/` | | r03 | quiet | 9743 | 30.48 / 26.84 | 0 | 20 s | GREEN (`ok`, 11 s) | 1 | 0 | 0 | 1 | `/Users/berni/labs-worktrees/r01-remeasure-evidence/runs/r03/` | | r04 | loaded | 9744 | 23.88 / 67.77 | 0 | 24 s | GREEN (`ok`, 13 s) | 1 | 0 | 0 | 1 | `/Users/berni/labs-worktrees/r01-remeasure-evidence/runs/r04/` | | r05 | quiet | 9745 | 54.38 / 42.47 | 0 | 17 s | GREEN (`ok`, 10 s) | 1 | 0 | 0 | 1 | `/Users/berni/labs-worktrees/r01-remeasure-evidence/runs/r05/` | | r06 | loaded | 9761 | 28.53 / 26.29 | 0 | 21 s | GREEN (`ok`, 13 s) | 1 | 0 | 0 | 1 | `/Users/berni/labs-worktrees/r01-remeasure-evidence/runs/r06/` | | r07 | quiet | 9747 | 18.46 / 17.54 | 0 | 19 s | GREEN (`ok`, 12 s) | 1 | 0 | 0 | 1 | `/Users/berni/labs-worktrees/r01-remeasure-evidence/runs/r07/` | | r08 | loaded | 9748 | 15.61 / 17.28 | 0 | 22 s | GREEN (`ok`, 13 s) | 1 | 0 | 0 | 1 | `/Users/berni/labs-worktrees/r01-remeasure-evidence/runs/r08/` | | r09 | quiet | 9749 | 15.24 / 13.84 | 0 | 12 s | GREEN (`ok`, 7 s) | 2 | 0 | 0 | 2 | `/Users/berni/labs-worktrees/r01-remeasure-evidence/runs/r09/` | | r10 | loaded | 9750 | 10.22 / 10.73 | 0 | 15 s | GREEN (`ok`, 9 s) | 1 | 0 | 0 | 1 | `/Users/berni/labs-worktrees/r01-remeasure-evidence/runs/r10/` | Quiet was 4/5 green and loaded was 5/5 green. Across the ten counted runs: - `deferred-start-catchup`: 12 - `deferred-start-catchup-failed`: 0 - `pattern-load-error`: 0 - stale-read lines: 12 - terminal `Error committing deferred` lines: 0 - target-step skip-print lines: 0 Every run used the same binary hash and measured head, passed the posture probe, started on a free port, and left its port free after PID-only teardown. ## Red classification The sole red, r01, is a new shape. The target step started after the preceding step passed in 20 seconds, reached `Await runtime idle for notebook regression...`, and remained in the target step until the unchanged 600-second harness bound returned rc 124. The durable target notebook space held 89 commits, 808 entities, and 1,220 revisions. Its notebook root had a real pattern identity. Its argument was a real notebook with exactly six note links, written in six post-creation patches at seqs 37, 45, 55, 63, 71, and 83. No seventh append was durable. The logs carried one `pattern-swap-setup-error`: > updated arguments do not match the candidate schema: parentNotebook: notes: > 0: parentNotebook: recursive schema validation made no progress They also carried seven `event-view-lag` deferral lines, reaching index 5, and zero `pattern-load-error` lines. This is not the earlier r01 signature: that required all seven appends durable, a live piece context, and a silently starved read. This store had six notes, an explicit setup error, and the step did not reach the final authority read. This is not the earlier r06/r09 signature: that required one keyless `pattern-load-error` followed by whole-piece unreadability while the durable pattern pointers remained real. This run had no `pattern-load-error`; its root was durable with a real pattern identity. The detailed classification is at `/Users/berni/labs-worktrees/r01-remeasure-evidence/runs/r01/classification.md`. The same directory carries `test.log`, `toolshed.log`, and the raw fresh store. ## Status of the earlier residues The read-side residues from the pre-#6292 gate are likely closed by intervening fixes, but the evidence is not symmetric: - r01 has a mechanism story: #6292 fixed the loss of a pre-watch `cid:` effect before `SpaceReplica` absorbed it, matching the read-side delivery loss that made r01 plausible. The old complete-store, live-context, silent-starvation signature did not occur in this campaign. - r06/r09 have absence-of-observation only. Their keyless-identity `pattern-load-error` mechanism was never root-caused, and the signature did not occur in this campaign. This campaign therefore supports “likely closed” for the earlier residues, not a proved common mechanism or an unconditional closure claim. ## Disposition No lift. The result was 9/10 green, so the STEP entry and its bound guard stay. No lift PR was opened from the campaign. The skip's current charge is the new store-incomplete setup-error shape. Its lift bar remains 10/10 green under the same quiet-and-loaded ON protocol. The campaign does not determine whether the seventh append was refused, dropped, or never issued, and it does not assign a root cause to the recursive-schema setup error. One excluded preflight is preserved at `/Users/berni/labs-worktrees/r01-remeasure-evidence/attempts/r06-port-9746-occupied/`. It found unrelated port holders and exited before launching toolshed. Counted r06 used free port 9761.