Brennan Benson
1d28b8882c
* chore(ratchets): count sync fs calls on Electron main and setInterval calls
Sync fs on main and fixed-cadence intervals that never settle were two of the
most common perf fix classes since September, and no gate counted them. Both
ratchets grandfather today's uses per file and fail when a file's count grows.
* chore(ratchets): count every sync fs export, src/shared and node:timers intervals
Review: the allow-list missed descriptor/copy APIs, src/shared helpers run on
main too, and setInterval imported from node:timers escaped the timer count.
* chore(ratchets): keep hosted setInterval calls beside a named timer import
Review: a named node:timers import zeroed window/globalThis calls too, a
destructured require was counted twice, and collector edits didn't select
the timer ratchet in check:changed.
* chore(ratchets): charge calls through an alias, not the alias binding
Decider ruling B on review finding C6: binding a plain alias of a gated call
counted as a use, so an alias refactor failed every collector-based ratchet.
Conditional/fallback aliases and exported bindings stay charged.
* chore(ratchets): count bare setInterval only when nothing local shadows it
Review: the timer ratchet kept its own narrower binding resolver, so alias
and namespace-destructure forms of an imported setInterval counted twice.
Module bindings are the shared collector's; a bare call is the global only
when the file declares no setInterval binding.
* chore(ratchets): resolve bare setInterval calls per scope
Decider ruling A: the file-wide shadow rule let an unrelated parameter or type
declaration named setInterval hide every global poll in the file. Each bare
call now resolves through a single-file TypeScript checker (noLib, noResolve)
and counts when it binds to nothing declared in the file.
* chore(ratchets): write the timer baseline exactly as --prune does
* chore(ratchets): charge global timer aliases and bracket access
CodeRabbit: `const schedule = setInterval; schedule(poll)` and
`window['setInterval'](poll)` counted zero. Count value references to the
global timer (per scope) instead of only direct calls.
* chore(ratchets): charge shorthand timer escapes; skip writes, typeof and JSX tags
Review: `{ setInterval }` of the global counted 0, while assignments to it,
parenthesized typeof probes and a <setInterval/> tag counted 1.
* chore(ratchets): charge compound assignments to the global timer
Review: `||=` and `??=` read setInterval before writing it, so only a
plain `=` target is write-only.
* chore(ratchets): regenerate baselines for files main moved
main's SSH reorganisation moved counted files under src/main/ssh and
src/main/managed-server; counts are unchanged except one sync fs call main
removed.
4.3 KiB
Performance regression checks
pnpm --silent audit:perf > performance-audit.json scans production src/ with
the existing app-store and buffer-concatenation rules. Warnings are advisory in this full inventory; tool/parser failures fail.
New findings from rules enabled in the changed-code configuration fail
pnpm check:code-quality:changed.
The app-store-performance/no-whole-hot-map-selector warning is advisory only:
it appears in this inventory and the performance-contracts artifact, and its
findings cannot fail the changed-code gate or other required checks. It catches
selectors returning the whole runtimeStatusByEnvironmentId map, which is
replaced on meaningful host observations. Select a stable derived answer inside
the selector (see paired-runtime-parking-capabilities.ts). Fleet consumers can
optionally silence the inventory warning with
// oxlint-disable-next-line app-store-performance/no-whole-hot-map-selector -- <why every host is needed>.
It resolves direct returns, module-scope named selectors and selector-local const
aliases/destructuring; it does not infer consumer lifetime or follow imported
helpers. Legitimate fleet consumers also match, so behavior-level render-isolation
assertions remain the safeguard against actual fanout regressions.
Tests, generated files, mobile/ and cloud/ are outside this source audit.
The ast-grep rule config/ast-grep/rules/no-repeated-collator.yml (pnpm lint:ast-grep,
an error everywhere) detects optioned localeCompare and Intl.Collator construction
inside inline sort/toSorted callbacks. Construct one collator outside the
callback, preserving locale, options and tie-breakers. If the locale changes at
runtime, reconstruct at the next sort or key the cache by locale. Bare comparisons
and standalone equality checks are allowed. There is no autofix or interprocedural
analysis: named comparators, aliases, custom methods and deferred callbacks need
manual review. A finding identifies repeated setup, not proof of visible lag.
pnpm test:perf:contracts runs the explicit selection in
vitest.performance.config.ts: SQLite statement reuse and schema parity, relay
filesystem concurrency, incremental Codex scans and reader lifetimes, pinned JSONL
readers and snapshots, linear tool-result pairing and attribution allocation,
tokenizer rejection, highlighting cache, queued cancellation, terminal backing-memory
retention and detector fixtures. Missing
listed files fail configuration loading. Tests run serially, without retries,
and inherit the full suite's setup and forced-GC support. This makes existing
regression coverage easy to run and attribute; it does not create new workload
coverage by itself.
Incremental Claude scans and persisted Claude cache checkpoints are also included.
.github/workflows/performance-contracts.yml runs daily and manually on Linux,
macOS and Windows, and on PRs changing this tooling or any listed contract file.
It uploads per-OS JSON test results, plus the source inventory once from Linux
because that scan is OS-independent. Its schedule starts after merge. Run the existing
test:e2e:terminal-perf:scale:report (guide) for rendered typing/frame budgets. Unit tests do not
measure SSH RTT, WSL scheduling or a packaged Electron renderer.
To extend coverage, select a production-path regression with an operation-count, identity, queue-admission or retained-memory oracle. Confirm it fails with the old behavior. Use controlled, counterbalanced benchmark samples for timings; avoid new machine-dependent millisecond gates in the normal unit suite. A green source scan and these contracts cannot establish that the whole app is fast.
Two count ratchets (counts per file may only fall) cover classes the audit only warns about:
pnpm check:main-sync-fs-ratchet counts synchronous node:fs calls (every *Sync export) in production src/main and src/shared
(config/main-sync-fs-baseline.txt), and pnpm check:recurring-timer-ratchet counts setInterval
calls in production src/main, src/renderer, src/shared and src/cli
(config/recurring-timer-baseline.txt). Both run in check:ratchets on every non-draft PR that runs static analysis. A new use either
moves off the hot path (async fs or a stat-fingerprint cache; a push signal, a visibility-gated or
backing-off timer) or raises its file's count with reviewer sign-off.