Skip to content

fix(g2-s1): sweep residual USD/generic-catalog text to kitchen/RUB legend#155

Merged
brownjuly2003-code merged 2 commits into
mainfrom
fix/g2-s1-usd-generic-sweep
Jul 5, 2026
Merged

fix(g2-s1): sweep residual USD/generic-catalog text to kitchen/RUB legend#155
brownjuly2003-code merged 2 commits into
mainfrom
fix/g2-s1-usd-generic-sweep

Conversation

@brownjuly2003-code

Copy link
Copy Markdown
Owner

Summary

Closes the M1-remainder + m10 findings from G2_AUDIT_REPORT.md (step S1 of new_plen_05_07_26.md, following #153):

  • contracts/entities/{product,user}.yaml: "in USD" -> "in RUB" (enum/unit already fixed in B1; this was free-text drift)
  • docker/postgres-source/init.sql: CDC seed currency USD->RUB, category electronics->kettles
  • scripts/nl_sql_eval/{warehouse,dataset}.py: generic Electronics/Home/Books demo catalog -> generator-spec.md §3 kitchen categories + RUB amounts; the one gold-SQL question tied to "Electronics" now asks about "kettles"
  • scripts/run_benchmark.py + tests/load/run_load_test.py: both carried a byte-for-byte duplicate of the pre-B1 generic seed (electronics/footwear/accessories, USD) — replaced with the same canonical §9 kitchen fixture already shipped in src/serving/backends/{duckdb,clickhouse}_backend.py, so these fixtures stop diverging from the real demo data
  • notebooks/01-agent-demo.ipynb: cells 6/8 hardcoded a $ prefix regardless of the returned currency; dropped it, then re-ran the order-lookup/revenue/NL-query cells against a local no-Docker duckdb-backend instance to replace the stale $-USD cached outputs with real, verified RUB values

M1's Currency-enum/events.py scope and EUR/GBP contract fixtures were already resolved in B1 — no test fixtures reference EUR/GBP as of this sweep.

Test plan

  • ruff check / ruff format --check — clean
  • mypy src/ --ignore-missing-imports — clean
  • pytest tests/unit/ tests/property/ — 1677 passed
  • pytest tests/integration/ — 261 passed, 52 skipped (no live services), 4 Docker-unavailable errors (pre-existing on this host); 2 timing-sensitive tests that failed under load in the full run (test_analytics_logging_is_non_blocking, test_batch_executes_items_concurrently_and_reports_wall_time) pass cleanly in isolation
  • scripts/export_openapi.py --check — no drift
  • scripts/check_schema_evolution.py — passed
  • CI (Kafka/ClickHouse/Postgres services + Docker)

…gend

Closes the M1-remainder + m10 findings from G2_AUDIT_REPORT.md:

- contracts/entities/{product,user}.yaml: "in USD" -> "in RUB" (the
  Currency enum/unit override already landed in the B1 fix; this was
  free-text description drift)
- docker/postgres-source/init.sql: CDC seed currency USD->RUB, category
  electronics->kettles (low-prio per m10, but trivial once touching the
  file)
- scripts/nl_sql_eval/warehouse.py + dataset.py: replace the generic
  Electronics/Home/Books demo catalog with generator-spec.md SS3 kitchen
  categories and RUB amounts; the one gold-SQL question tied to the old
  "Electronics" category text now asks about "kettles"
- scripts/run_benchmark.py + tests/load/run_load_test.py: both had a
  byte-for-byte duplicate of the pre-B1 generic products_current/
  orders_v2/users_enriched seed (electronics/footwear/accessories, USD).
  Replaced with the same canonical SS9 kitchen fixture already shipped in
  src/serving/backends/{duckdb,clickhouse}_backend.py, so the benchmark
  and load-test fixtures stop diverging from the real demo data.
- notebooks/01-agent-demo.ipynb: cells 6/8 hardcoded a "$" prefix ahead
  of the currency code regardless of what the API returned; dropped it.
  Re-ran the notebook's order-lookup/revenue/NL-query cells against a
  local no-Docker duckdb-backend instance (SERVING_BACKEND=duckdb,
  AGENTFLOW_SEED_ON_BOOT=true) to replace the stale $-USD cached outputs
  with real, verified RUB values from the canonical demo dataset.

M1's Currency-enum/events.py scope and the EUR/GBP contract fixtures
were already resolved in the B1 fix; no further test fixtures reference
EUR/GBP as of this sweep.

Verify: ruff + ruff format + mypy clean; unit+property 1677 passed;
integration 261 passed + 2 flaky timing tests confirmed passing in
isolation (test_analytics_logging_is_non_blocking,
test_batch_executes_items_concurrently_and_reports_wall_time) + 52
self-skip (no live Kafka/ClickHouse/Postgres) + 4 errors from Docker
being unavailable on this Windows host (pre-existing, not a
regression); openapi --check and check_schema_evolution.py both clean.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
@github-actions

github-actions Bot commented Jul 5, 2026

Copy link
Copy Markdown

DORA Metrics

  • Window: last 30 days
  • Branch: main
  • Deployment frequency: 87 total / 20.3 per week
  • Lead time for changes: avg 0.55h / median 0.0h
  • Change failure rate: 58.62% (51/87)
  • MTTR: n/a across 0 incident(s)

MTTR note: No failed mainline CI runs in the selected window.

@brownjuly2003-code brownjuly2003-code merged commit dc23318 into main Jul 5, 2026
20 checks passed
@brownjuly2003-code brownjuly2003-code deleted the fix/g2-s1-usd-generic-sweep branch July 5, 2026 20:21
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants