You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -52,7 +52,7 @@ docker compose up --build
52
52
53
53
The local worker defaults to `http://localhost:8797`.
54
54
55
-
Current local qualification note: Lighthouse is greenand the worker throughput gate passes, but the browser 100k-row p95 is still `2076.14 ms` versus the required `<= 1000 ms`.
55
+
Current local qualification note: Lighthouse is green, browser p95 is `377.26 ms` for 100k rows, and the worker throughput gate passes. Remaining blockers are Miller parity completion, browser Miller-WASM evidence, hosted deployment, and final Section 21 closure.
Copy file name to clipboardExpand all lines: docs/qc/APPENDIX_B_REPORT.md
+3-6Lines changed: 3 additions & 6 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -9,8 +9,8 @@ Verifier: Codex
9
9
10
10
Counts:
11
11
Total checks: 34
12
-
Passed: 18
13
-
Failed: 1
12
+
Passed: 19
13
+
Failed: 0
14
14
Blocked: 15
15
15
16
16
Passed:
@@ -31,11 +31,9 @@ Passed:
31
31
- 21.18 A1 sample outputs are recorded in this appendix for fixture-backed evidence.
32
32
- 21.18 A2 worker escalation prompt covered by local unit tests and UI path.
33
33
- 21.12 Worker native DuckDB benchmark clears the 100M-row threshold: `1057.87 ms` for `generate_series(1, 100000000)`.
34
+
- 21.12 Browser 100k-row p95 now clears the target: `377.26 ms` in the production-preview run recorded at `docs/qc/benchmarks/browser-duckdb.json`.
34
35
- Local Lighthouse preview scores clear the handoff gate: `performance=100`, `accessibility=100`, `best-practices=100`, and `seo=100`.
35
36
36
-
Failures:
37
-
- 21.12 Browser 100k-row p95 still misses the target: `2076.14 ms` in the production-preview run recorded at `docs/qc/benchmarks/browser-duckdb.json`, versus the required `<= 1000 ms`.
38
-
39
37
Acceptance evidence:
40
38
- Ecommerce CSV fixture aggregates to `books=42.5`, `electronics=129.99`, and `home=77.1` with one paid row per category.
41
39
- Access-log JSONL fixture jq-style query returns request `r2` for path `/cart` with duration `983`.
@@ -47,7 +45,6 @@ Blocked:
47
45
Browser DuckDB-WASM is now exercised by `pnpm --filter @csvshape/web smoke:duckdb`, which loads the ecommerce CSV sample, applies `filter -> stats1`, shows `Engine: DuckDB-WASM`, and captures `docs/qc/screenshots/duckdb-wasm-preview.png`.
48
46
Worker-native DuckDB is packaged via `@duckdb/node-api`, reports `duckdbNative: true` on `/health`, and executes inline CSV/JSONL SQL plus Parquet export in local tests.
49
47
Worker-native Miller now has a repeatable smoke run via `pnpm --filter @csvshape/worker smoke:mlr`, which returns `engine=mlr-native`, `rowCount=3`, and CSV output for the paid-order ecommerce subset in `docs/qc/benchmarks/native-mlr-smoke.json`.
50
-
- 21.12 Browser auto mode now routes simple single-source chains to the faster TypeScript preview path, and the benchmark now runs against a production preview build instead of the dev server, but the browser p95 is still above target.
51
48
- 21.13 Privacy evidence is now local-only rather than hosted.
52
49
`pnpm audit:privacy` produces `docs/qc/benchmarks/browser-privacy.json`, which currently shows no worker calls, no cross-origin calls, and no browser storage writes during a standard browser-side sample transform.
53
50
Worker `/health` and `/v1/run` responses continue to expose `artifactTtlSeconds=900` for retention handling, and worker tests now assert `Cache-Control: no-store` on those responses.
Copy file name to clipboardExpand all lines: docs/qc/SECTION_21_STATUS.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -34,7 +34,7 @@ It is not yet a qualified release record.
34
34
Worker-native DuckDB execution is covered by local tests, and worker-native Miller now has a repeatable local smoke run via `pnpm --filter @csvshape/worker smoke:mlr` with artifact output in `docs/qc/benchmarks/native-mlr-smoke.json`.
35
35
Browser Miller-WASM evidence and per-verb Miller parity evidence are still missing.
36
36
-[ ] 21.12 Performance evidence for browser p95 and worker throughput.
37
-
Browser 100k-row p95 currently measures `2076.14 ms` via the production-preview benchmark in `docs/qc/benchmarks/browser-duckdb.json`, which still misses the `<= 1000 ms` gate.
37
+
Browser 100k-row p95 now measures `377.26 ms` via the production-preview benchmark in `docs/qc/benchmarks/browser-duckdb.json`, which clears the `<= 1000 ms` gate.
38
38
Worker native DuckDB clears the 100M-row threshold at `1057.87 ms` via `docs/qc/benchmarks/worker-duckdb.json`.
39
39
Local Lighthouse preview scores are `performance=100`, `accessibility=100`, `best-practices=100`, and `seo=100` via `docs/qc/benchmarks/lighthouse-summary.json`.
40
40
-[ ] 21.13 Privacy proof for browser-first processing and worker retention TTL handling.
0 commit comments