cold-store-reaper(test): suite takes 162s for 80 tests — every test pays ~1.5s of fresh-manifest schema creation in setUp

Finding

The cold-store-reaper test suite takes 162 s for 80 tests, and the cost is not the tests — it is per-test setup: every test creates a fresh cold-store home, and creating the manifest schema costs ~1.3–1.7 s each time.

Verified this run (main 37418e56, sandbox VM):

python3 -m unittest discover -s pkgs/cold-store-reaper/tests → Ran 80 tests in 162.238s (OK, 1 skipped).
Per-test timing: slowest single test 6.2 s, only 13 of 80 under 2 s, test bodies near-instant (e.g. test_auth_url_shape body: 0.01 s; its setUp: 1.51 s).
Isolation: ColdStore(home).init() on a fresh home = 1.67 s; second init() on the same home = 0.01 s. cProfile of a fresh init(): 1.30 s total, of which 1.295 s is a single call — sqlite3.Connection.executescript inside ColdStore.db() (cold_store.py:369), running schema.sql to create the manifest.
Every test class derives from ColdStoreTestBase.setUp (tests/test_cold_store.py:232), which builds a brand-new TemporaryDirectory home and calls store.init() — so the suite pays the fresh-schema cost ~80 times (plus stacked fake Drive/S3/D1 servers in subclasses). The turso-export and drill tests pay it again per test through their own fixtures.

The absolute executescript cost varies with filesystem fsync speed (this VM is slow; a fast CI disk will be quicker), but the structure — full schema creation per test instead of once per suite — dominates the runtime anywhere and grows linearly with the test count.

Why it matters

A 2–3 minute suite is a suite people stop running locally — which compounds #1427 (no CI runs these tests at all): the package's only safety net is both unwired and slow. It also slows every future CI wiring: a path-filtered workflow (per #1427) that takes ~3 minutes for one small package will pressure people to skip it. Backup-integrity code deserves a suite that runs in seconds.

Acceptance criteria

The suite's wall-clock drops by at least ~5x on this VM (162 s → under ~35 s) with all 80 tests still passing and none deleted or weakened.
Fresh-schema creation happens once per suite/run (e.g. a template manifest built once and copied per test, or an equivalent shared fixture), while each test still gets an isolated manifest it can mutate freely.
Tests that specifically exercise first-run schema creation/migrations (fresh-home init, pre-v4 manifest migration) keep a genuinely fresh home — they are the exception, not the default fixture.
The measurement is repeated and recorded on the PR (before/after wall-clock for the full discover run).

CODE-519

Filed this run by the new-issue-creator (lane A). Measured on main 37418e56: full discover run 162.238s for 80 tests; per-test timing shows bodies near-instant and setUp dominant (test_auth_url_shape: setUp 1.51s, body 0.01s); cProfile of a fresh ColdStore.init() puts 1.295s of 1.30s in the single executescript call in ColdStore.db() (cold_store.py:369). Companion to #1427 (no CI runs the suite).

Session / ID: cron new-issue-creator-code-a run 2026-10-27 10:05 CDT
Found via: filed this run
Task: filed the issue; queued a triage packet for routing
Expectation: effort/m — done when the suite runs ≥5x faster via once-per-suite schema creation, with fresh-home tests (init/migrations) keeping real fresh homes
Links: packet enqueued to repo-health-code-pilot this run

🤖 Agent note · picked up

Agent: queue worker (repo-worker-code), Lane A shift
Session / ID: queue task 1438
Found via: queue claim
Task: Triage issue cold-store-reaper(test): suite takes 162s for 80 tests — every test pays ~1.5s of fresh-manifest schema creation in setUp #1435: re-verify the per-test fresh-schema cost claim against current main (structure + a fresh-init re-measurement on the laptop), post a verdict, and if confirmed enqueue the fixture small_fix.
Expectation: effort/s — done looks like a verdict on cold-store-reaper(test): suite takes 162s for 80 tests — every test pays ~1.5s of fresh-manifest schema creation in setUp #1435 and, if confirmed, a queued small_fix follow-up.
Links: Issue cold-store-reaper(test): suite takes 162s for 80 tests — every test pays ~1.5s of fresh-manifest schema creation in setUp #1435 · Companion ci(cold-store-reaper): no workflow runs the package's test suites — 2,700+ lines of tests execute only when a human remembers #1427

🤖 Agent note · parked

Agent: queue worker (repo-worker-code), lane B shift
Session / ID: queue task 1492
Found via: queue claim
Task: PR test(cold-store-reaper): template-home fixture — suite 87.2s → 10.9s (#1435) #3254 opened for cold-store-reaper(test): suite takes 162s for 80 tests — every test pays ~1.5s of fresh-manifest schema creation in setUp #1435 (template-home fixture). Parked 20m on CI: python: unittest suites SUCCESS; worker: local proof still QUEUED after a 5-minute poll. Codex review run failed on Codex usage limits with zero finding threads (verified: no review comments on the PR), CodeRabbit rate-limited — no findings to disposition.
Expectation: on resume — if worker: local proof is green, merge PR test(cold-store-reaper): template-home fixture — suite 87.2s → 10.9s (#1435) #3254 (issue closes), then complete task 1492 as fixed_merged. Laptop evidence: suite 87.192s → 10.875s/10.492s, 104 tests OK (skipped=1) both runs.
Links: PR test(cold-store-reaper): template-home fixture — suite 87.2s → 10.9s (#1435) #3254 · branch queue/1492-1435-template-home-fixture @091224592

agent-note:v1

event: picked_up
task_id: 1492
queue: repo-health-code-pilot
worker: repo-worker-code (lane D)
stage: verified
pr: https://github.com/developing-today/code/pull/3254
note: >
  Resume: run 41253284293 now COMPLETE — python: unittest suites PASS and
  worker: local proof PASS on head 091224592, so the park's resume
  condition is met. PR mergeable_state=clean, 0 review threads. Fresh
  Codex re-review requested; merging once it lands clean (or quota-fails
  with zero findings, per the prior park note).

Merge pull request #3254 from developing-today/queue/1492-1435-templa…

agent-note:v1

event: done
task_id: 1492
queue: repo-health-code-pilot
worker: repo-worker-code (lane D)
stage: bookkeeping_complete
pr: https://github.com/developing-today/code/pull/3254
merge_sha: 7abf085055c51fa03c71ce9d749eff7b09988260
note: >
  PR 3254 MERGED and #1435 closed. Run 41253284293 completed green
  (python: unittest suites PASS, worker: local proof PASS) and the fresh
  Codex re-review on head 091224592 reported "Didn't find any major
  issues" — 0 review threads. Task completed fixed_merged.

Navigation Menu

Repository navigation

cold-store-reaper(test): suite takes 162s for 80 tests — every test pays ~1.5s of fresh-manifest schema creation in setUp #1435

Finding

Why it matters

Acceptance criteria

18 remaining items

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

cold-store-reaper(test): suite takes 162s for 80 tests — every test pays ~1.5s of fresh-manifest schema creation in setUp #1435

Description

Finding

Why it matters

Acceptance criteria

Activity

developing-today commented on Oct 27, 2026

dezren39 commented on Oct 27, 2026

dezren39 commented on Oct 27, 2026

18 remaining items

dezren39 commented on Oct 31, 2026

dezren39 commented on Nov 1, 2026

dezren39 commented on Nov 1, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions