easy-stack

Testing and evidence

Five focused suites. One non-duplicated catalog.

Every case has one stable ID and one suite owner. Behavioral acceptance scenarios cross lower-level feature contracts without duplicating their definitions, and runtime replays never inflate the unique count.

Explore each suite

16 cases · U001–U016

Unit

Small, isolated contracts for Stack state, validation, references, and source preparation.

 24 cases · F001–F024

Functional

Focused LIFO, control hand-off, active insertion, build-parity, and Playground feature contracts.

 8 cases · B001–B008

Behavioral

Composite consumer outcomes plus real-Chrome journeys through visible Playground controls and editor recovery.

 29 cases · I001–I029

Integration

Package entries, metadata, docs, workflows, server delivery, and executable source replay.

 17 cases · R001–R017

Regression

Named guards for recovered defects, path safety, source injection, sandboxing, and Worker lifetime.

How the counts work

EvidenceCountRule
Unique cases94One stable ID and contract key per catalog case, counted once.
Runtime executions727Expected replays on a successful main push across the CI matrix, coverage runtimes, Behavioral Chrome, Regression Chrome, and Pages verification.
Gated job executions10Six OS/runtime matrix jobs plus coverage, package, benchmark, and curated-site gates.

Running a case again on another Node version, operating system, or browser is an execution—not a newly claimed test. Coverage, package smoke, and deployment are gates rather than case definitions.

Native V8 coverage

easy-stack works with bundlers and without a bundler. Bundlers resolve the bare package import normally; native browser ESM uses an explicit import map with no build or transpilation step. The same 26-case host-neutral inventory imports easy-stack in Node and real Chrome, and a separate gate compares the ordered result inventories. Statements, branches, functions, and lines must each reach 100%.

npm test
npm run test:unit
npm run test:functional
npm run test:behavioral
npm run test:behavioral:chrome
npm run test:integration
npm run test:regression
npm run test:regression:chrome
npm run coverage
npm run benchmark

CI and publication gates

  1. Run the host-neutral portions of all five suites on exact Node 22.13 and current Node 24 across Linux, macOS, and Windows.
  2. Measure 100% native V8 coverage in Node and Chrome.
  3. Run five real-Chrome Behavioral journeys, including the observable native no-bundler import-map scenario, and two Regression guards through the opaque Playground frame and classic Worker.
  4. Pack, inspect, install, and prove ESM/CommonJS constructor identity in a clean consumer project.
  5. Compare the exact 2.0.0 release with 2.1.0 and upload checksum-validated benchmark evidence.
  6. Deploy the curated Pages artifact only after the exact main-branch CI commit succeeds.
  7. Accept only a bare numeric release tag matching package.json, then publish to npm through GitHub OIDC trusted publishing.
Runtime dependency count: zero.

vanilla-test@2.1.1 is pinned development tooling and never enters the published dependency graph.

Open the first catalogUnit suite