0 Fork 0

mirror of https://github.com/PR0M3TH3AN/bitgate.git synced 2026-10-27 08:31:00 +00:00

Files

T

thePR0M3TH3AN 7f70a64d61 test: remove flakiness from the performance and widget suites

A full-suite run failed intermittently, roughly one in six. Two causes, both
mine, both in test code rather than the library.

The performance fixture's setup repeated the same 100 mute-list writes 50 times
over with identical data — 5,000 calls where 100 give the same fixture — which
made construction expensive enough that individual tests occasionally exceeded
their timeout under load. The failures looked like assertion failures on
deterministic checks, which is what made them confusing. The fixture's shape
(5,000 targets, 500 authors, 100 muters per author, multiple categories and
contributors) is now asserted directly so it cannot silently degrade again.

The wall-clock assertion is gone. It measured the machine as much as the code —
85ms standalone versus 1606ms under vitest on the same commit — so it was flaky
rather than informative. Replaced with structural assertions that the snapshot
and its record maps are materialized once per batch, which is precisely the
regression that made a 5,000-target pass take ~47 seconds, and is deterministic.

Widget tests polled with vi.waitFor and its ~1s timeout. The async elements now
expose the in-flight promise as `pending`, so tests await completion directly.
That is also useful to hosts wanting to disable surrounding UI mid-action.

Six consecutive full-suite runs, 690 tests each, no failures.

2026-08-09 19:36:30 -04:00

6.1 KiB

Raw Blame History

Changelog

Notable changes to this repository. The package version and the wire schema version move independently: a breaking package release does not imply a new event schema, and a new schema does not require a major package bump unless the public API changes with it.

[Unreleased]

Added

Core decision model. Four independent dimensions (ranking, visibility, interaction, transaction) with severity ladders and commutative max-merge composition.
Stable reason identifiers (REASON_IDS) as part of the public contract. Renaming or removing one is a breaking change.
Structured evidence, redacted by default and merged across an object's targets, so a multi-target object explains itself with everything that contributed.
Administrative-state reduction with capability filtering at merge time, community-source marking, and protected-actor stripping.
Snapshot fingerprints for cache invalidation and mismatch records.
Application adapter contract (createApplicationAdapter, evaluateObject, collectTargets).
Codecs: canonical v1 on kind:30078, NIP-56 reports, NIP-51 mute lists, legacy administrative lists, bech32/NIP-19, replaceable selection with a deterministic same-timestamp tiebreak, injected signature verification.
Runtime: six stores, transport orchestration, chunked report subscriptions, viewer lifecycle, diagnostics, idempotent teardown.
Commands with local capability checks, protected-target refusal, partial relay acceptance, and stable error codes.
Characterization corpus: 34 generated fixtures covering the behaviors the migration plan's Phase 0 requires, plus a conformance suite that evaluates them.
Commerce proof-of-fit (examples/commerce) demonstrating the engine outside a video application using only the public API.
Decision caching keyed by (profile, target) with targeted invalidation: report and mute changes invalidate only the targets they touch, while administrative, trust, policy, and viewer changes drop the whole cache. Cached decisions are deeply frozen so a consumer cannot corrupt later reads.
Performance fixture (tests/performance/) over 5,000 targets, 500 authors, and 100 trusted muters per author, asserting no network access, bounded subscription counts, and targeted invalidation.
Security requirements suite (tests/security/) covering the enumerated requirements testable at this layer.
@nostr-governance/widget: framework-agnostic custom elements. Viewer surfaces (governance-veil, governance-report, governance-status) and moderator surfaces (governance-capabilities, governance-action, governance-admin-panel), plus a runnable demo page. Elements render decisions and issue commands; they compute no policy.
Runtime wiring: storage persistence and hydration, signature verification in the load path, trusted mute-list subscriptions, community source resolution, root-policy application, import/export, and capability queries.
GPL-3.0-or-later LICENSE, README, and an integration guide.

Fixed

Per-package types fields pointed at paths the build never produced, so consumers received no type declarations at all. Declarations now emit per package and resolve.
reduceAdminState accepted malformed target identifiers, because getTargetKey formats anything it is given.
community-user-deny fired for any contributor rather than only for community-curated sources, making a direct moderator action indistinguishable from a federated one.
composeDecisions kept only the first decision's evidence, so an object governed by several targets lost the evidence that drove its verdict.
OverrideStore.clear() shadowed the inherited listener clear(), leaking subscribers on teardown. Now clearOverrides().
GovernanceAdminStore emitted only when reduced denial state changed, so revoking a role that happened to deny nobody announced nothing — leaving capability-gated UI and the decision cache stale. It now also emits on authority changes.
ReportStore.ingest required a target field it never read, since the target is identified by the key argument.
The performance fixture was intermittently failing. Its setup repeated the same 100 mute-list writes 50 times with identical data, making construction 50x more expensive than the fixture it describes and pushing individual tests past their timeout under load. The fixture's shape is now asserted so it cannot silently degrade again.
The performance suite's wall-clock assertion was replaced with structural ones. A timing budget measured the machine as much as the code — it varied roughly 20x between idle and loaded runs here — so it now asserts that state is materialized once per batch, which is the actual regression that made a 5,000-target pass take ~47 seconds.
Evaluation was quadratic in state size: both the snapshot and its fingerprint were rebuilt for every target, so each call walked every report and mute list. A 5,000-target pass took ~47s. The runtime now memoizes the snapshot and its fingerprint until a store changes, and evaluateMany computes the fingerprint once for the batch. The same pass now takes ~85ms cold and ~6ms warm.
npm audit reported 5 vulnerabilities (1 critical, 2 high, 2 moderate), all from [email protected]'s dependency tree (vite → esbuild, postcss → nanoid). Dev-only and never shipped, since packages publish src and dist alone. Upgrading to vitest@3 clears all five with no test changes.

Changed

The evaluator no longer carries application thresholds. The previous invented defaults (5/3/10/5) are gone; applications supply category thresholds through a policy definition. The characterized reference values live in @nostr-governance/bitvid-compat as one application's profile.
createAuthorityState takes an options object and always grants the root every capability and protection.
Evaluation takes a GovernanceSnapshot and a ViewerState rather than a flat state object, and reads an injected clock instead of wall time.