changelog.md

0.2.1

• fix: ship dist/multihook.d.ts. dist/interop/tapable-interop.d.ts re-exports MultiHook from ../multihook, and 0.2.0 excluded that declaration from the build, so a consumer typechecking without skipLibCheck got TS2307: Cannot find module '../multihook'. The runtime was never affected: Interop.MultiHook has been in the bundle throughout
• fix: the declarations typecheck under TypeScript's nodenext resolution. Relative imports in src had no file extension, so every shipped .d.ts failed with TS2834 and the entry point exported nothing to a nodenext consumer. 0.1.2 was unaffected because its declarations were rolled into one file; 0.2.0 introduced it
• test: every relative import in a shipped declaration must carry an extension and resolve to a shipped file
• docs: the readme said Multihook is not exported. It is, as Interop.MultiHook
• docs: say what declines. Only undefined does: bail treats null as an answer, and waterfall treats null and '' as values
• task: the author and copyright holder are Steven Spungin, in package.json, license.txt and the licence header of every shipped file. The twelvetone keyword is removed. The licence is still MIT

0.2.0

• breaking: bail now runs in series for async hooks, matching the sync bail, the readme, and tapable's AsyncSeriesBailHook. It previously raced every tap through Promise.any
• feat: add the race topology: every tap starts, the first to resolve wins. This is the behaviour the async bail used to have, and it matches tapable's AsyncParallelBailHook
• feat: expose Interop.AsyncParallelBailHook, which had no equivalent before
• fix: bail returns the value that caused it to bail. It short-circuited correctly but discarded the value, so call answered undefined whether a tap answered or none did
• fix: waterfall passes the accumulated value through a tap that returns nothing, matching tapable. A silent tap previously handed the literal undefined to the next tap
• fix: vite build completes. rollupTypes aborted it on "Unable to follow symbol for Partial", before the UMD pass, so dist/index.umd.cjs was never emitted
• test: three tests asserted the old behaviour rather than the contract; corrected, with notes. Verified against tapable 2.3.3 directly
• fix: race no longer logs an error when no tap answers. Promise.any rejects once every tap has rejected, and a tap returning undefined is rejected internally, so the normal "nobody answered" outcome arrived as an AggregateError and was logged. A real exception in a tap is still reported
• fix: the type documentation had the second and third template parameters the wrong way round. The signature is PARAMS, TAPDATA, CONTEXT; the comment said context was second, and context.spec.ts was written from the comment. Nothing caught it because the tests were never typechecked
• task: bump vite to ^5.4.21 and vitest to ^1.6.1 within their majors, clearing 13 of 17 advisories including one critical. Both are devDependencies; this package has no runtime dependencies at all, so none of them ever reached a consumer. The remaining four need vite 6.4.2+ and vitest 3.2.6+
• task: tests no longer print. Four vi.spyOn(console, 'error') calls recorded the call but had no mock implementation, so they called through and every deliberately provoked exception was printed
• fix: samples/example.cjs runs. It had top-level await in a CommonJS file, so node samples/example.cjs threw a SyntaxError, including in the published 0.1.2. Its test had imported the file through vitest, which transforms it and supports top-level await, so the test environment was not the user's and could not see the defect. The test now runs it under plain node
• fix: stop shipping dist/hookmap.d.ts and dist/multihook.d.ts. Neither module is exported from the entry point and neither is in the bundle, so those declarations described an API with no runtime behind it. The readme's feature table now says so rather than claiming both are available
• feat: ship src, test and sourcemaps. The library build mangles identifiers and strips comments, so dist/index.js was unreadable (every function a single letter), and a consumer's stack trace pointed into it with no way back. There is nothing in this package worth withholding, and the cost is a tarball of 53 kB instead of 17 kB. The tests ship for the same reason: they are the evidence for what this package claims, including its conformance against tapable 2.3.3
• fix: sourcemaps are excluded from the licence-header step. A .map is JSON, and prepending a comment block to it makes it invalid, so every debugger would have silently fallen back to the mangled bundle
• feat: add an exports map. require now resolves to the UMD build, so CommonJS consumers work on Node below 22.12, where require of an ES module fails outright. Deep paths under dist/ still resolve, so nothing that reached for the UMD build by hand breaks

0.1.1

• task: Upgrade dependencies

0.1.0

• breaking: Change parameter order of TAPDATA and CONTEXT
• fix: Fix typo in function name
• docs: Update docs

0.0.6

• fix: Fix types

0.0.5

• task: Rollup types

0.0.4

• task: Add dts plugin
• task: Upgrade dependencies

0.0.3

• docs: Update readme

0.0.2

• task: Upgrade dependencies
• docs: Update readme

0.0.1

• fix: restore prepack
• docs: Update readme

0.0.1-beta.7

• fix: disable prepack
• docs: Update readme

0.0.1-beta.6

• fix: Update types
• task: Update vitest, vitenode

0.0.1-beta.5

• fix: Fix interop types
• docs: Update readme
• docs: Update examples

0.0.1-beta.4

• task: Refactor types
• docs: Update parameter, usage, and error handling

0.0.1-beta.3

• task: Update docs
• fix: Remove console.log
• docs: Add usage overview
• docs: Add more interop info

0.0.1-beta.2

• task: Publish initial project
• task: Monkeypatch Tapable unit tests
• feat: Add Tapable interop

0.0.1-beta.1

• task: Publish initial project shell