hygiene(tick-history): 2026-05-01T20:35Z — substrate-densification tick

docs/hygiene-history/ticks/2026/05/01/2035Z.md

@@ -0,0 +1 @@
+| 2026-05-01T20:35:00Z | opus-4-7 / autonomous-loop tick | 98fc7424 | Substrate-densification tick — three landings on PR #1153 + new PR #1155. **REFRESH WORLD MODEL via `bun tools/github/poll-pr-gate-batch.ts 1153 1154`** (eating own dogfood from prior tick): #1153 BLOCKED, 1 unresolved thread (Codex P2); #1154 BLOCKED, wait-ci. (1) **Codex P2 fix** — Codex flagged that `--all-open` silently truncates at 1000 PRs because `gh pr list --limit 1000` has no pagination. Fixed by replacing with `gh api --paginate /repos/{owner}/{repo}/pulls?state=open&per_page=100`. Live-tested on 27-PR LFG queue. Commit `275458e`. (2) **DST tests** — Aaron 2026-05-01 *"and you can DST it"*. New `tools/github/poll-pr-gate-batch.test.ts`: 10 tests in 152ms, zero `gh` spawns, zero network, zero clock dependencies. Refactored `pollAllBounded` to take injectable `pollFn` (defaulted to `pollOne`). Coverage: empty input, multi-axis counting, actionable-exclusion contract, per-PR-prefixed warnings, input-order preservation, out-of-order resolution under bounded concurrency (peak in-flight assertion), worker-count clamping, error propagation without exception cascade. Worked-example proof for the dynamic-bash memory file's load-bearing claim *"DST is achievable in TS."* Commit `5848fa1`. (3) **SQLSharp external anchor memory** — Aaron 2026-05-01 four-message clarification (*"`../SQLSharp` is a decent example... best bun/ts practices... super strongly typed... not js in ts failure mode"*) pointed at `../SQLSharp` as the bun/ts pattern anchor (distinct from DST anchor). Pattern: `Dependencies` interface for every external call (`typeof X`-keyed) + `defaultDependencies` const + (args, options, deps?) call shape + tests pass synthetic deps. New memory file `feedback_ts_dependencies_as_interface_di_pattern_sqlsharp_anchor_aaron_2026_05_01.md` with worked-example gap analysis on poll-pr-gate-batch.ts (only `pollFn` injectable; full pattern would extract `spawnSync`/`process.exit`/`process.stdout`/`new Date()` to BatchPollDependencies interface — follow-up refactor target). PR #1155 opened on `otto/sqlsharp-bun-ts-pattern-anchor-2026-05-01` (cherry-picked clean from misplaced commit on poll-pr-gate-batch branch). Cron 98fc7424 healthy. | [PR #1153 +2 commits (Codex P2 fix + DST tests); PR #1155 OPENED with SQLSharp-pattern memory; both auto-merge armed; world-model refresh validated the new tooling on its own substrate] | The session arc this tick: Aaron's *"DST it"* + *"SQLSharp is a decent example"* are coupled corrections. The DST tests prove the dynamic-bash memo's load-bearing claim (TS unlocks DST grade-A); the SQLSharp memo names the form-factor target the tools should write toward (super-strongly-typed DI pattern, not js-in-ts). The two together make poll-pr-gate-batch.ts an iteration-1 worked-example: DST boundary in place (pollFn injectable), but full pattern not yet applied (spawnSync etc still hardcoded). Future refactor follows SQLSharp `formatRepoDependencies` shape. The carved blade from #1155: *"Super-strongly-typed TS, not js-in-ts. Every external call surface in an interface. The pattern IS the DST achievability vehicle."* |