# @systemfsoftware/oxlint-plugin ## 4.0.4 ### Patch Changes - Bump the stryker catalog to pull in the latest patch releases of the stryker-js ecosystem. ## 4.0.3 ### Patch Changes - Dependencies resolve to Effect 4.0.0-rc.116 and Vitest 5. ## 4.0.2 ### Patch Changes - The recommended oxlint set now carries `workflow-variant-constructed` and `runtime-construction-placement` at `error`, and the make-keyed workflow rules recognize the `Workflow.total` and `Workflow.andThen` constructors as lawful workflow construction. ## 4.0.1 ### Patch Changes - Releases the workspace so its published versions track the shared dependency graph this change moves. ## 4.0.0 ### Major Changes - The `no-domain-branching-density` rule is removed. It is gone from the plugin's recommended set, so a project that extends the recommended configuration needs no change; a project that names the rule directly must delete that configuration entry, because an id no rule answers to is reported as unknown and enforces nothing. Per-function cyclomatic complexity is now measured by the linter's built-in `complexity` rule. The complete preset configures it; a project adopting the plugin alone does not get it. ## 3.0.0 ### Major Changes - Two rules are removed: `no-escaping-module-state` and `no-effect-service`. Both are gone from the shared base config, so a project extending it needs no change; a project naming either rule directly must delete that entry, and a project relying on either for enforcement should know neither was enforcing much. `no-escaping-module-state` refused a module-scope `Map`, `Set`, `Ref`, `Deferred`, `Queue` or semaphore. It read one file's syntax, so it decided the question only for the spelling it happened to see: the same primitive reached a consumer unreported through an object or array literal, a class static field, a `globalThis` assignment, a factory return, a destructured or computed member, or an immediately-invoked function. Its type-annotation arm was keyed on a closed list of type names, which meant every primitive absent from the list — the STM and concurrent families among them — passed, and every future one would need adding by hand. It also refused the shape its own message prescribed, reporting a module-private binding that was never exported at all. `no-effect-service` refused `Effect.Service`. Effect v4 exports no `Service` from the `Effect` module, so every use of it is already a type error naming the exact expression and offering the replacement — the rule restated a compiler diagnostic, and restated it only for a direct member access on a named import, missing a local alias, a computed key, a destructure, a namespace import, a cast, a re-export and a default import. Declare a service as `class X extends Context.Service()('id')`, adding a `make` option when the module owns its single construction. Build a coordination primitive inside the function that yields the surface and export the operations rather than the interior — `withLock(key, effect)`, `joinInFlight(key, effect)`, `read()` — so a caller depends on behaviour and swapping the interior or adding backpressure breaks nobody. ### Minor Changes - The core regime keys on the `Workflow.make` boundary; the complement gains a complexity ceiling. - `workflow-match-exhaustive` no longer reads the filename: the gate is the make callee boundary (import binding + member `make` + argument containment, module-scope references followed, shadow-correct). Identical dispatch outside a make body produces no diagnostic. - New `make-body-purity`: references inside make bodies resolve only to parameters, const locals, and audited-pure imports; control flow is banned with the one first-statement converging guard; unclassifiable references report honestly as unresolvable rather than passing. Test files are exempt — fixtures exercise decisions without the production regime binding them. - New `no-domain-branching-density` in core: per-function McCabe CC outside make bodies, ceiling 17 — the lowest measured value the tree passes with zero waivers (max measured 17; the 15 functions over 10 are the recorded extraction backlog, not retrofit targets). - Both make-boundary rules fix classifier defects their first workspace run exposed: builtin globals with empty defs take the named-global triage, and `as const` type subtrees leave the value-reference walk. - cut over to effect v4 (4.0.0-rc.108): public surface derives from effect types; peers flip effect ^3→^4 ### Patch Changes - New version is published through npm trusted publishing, so it carries a provenance attestation you can verify. - Peer Effect requirement advances to 4.0.0-rc.112. No API changes. ## 2.1.3 ### Patch Changes - Re-published against the oxc-based instrumenter: workspace dependency ranges move to the new instrumenter major; no package's own behavior changes in this release. ## 2.1.2 ### Patch Changes - Peer Effect requirement advances to 4.0.0-rc.112. No API changes. ## 2.1.1 ### Patch Changes - `ban-classes` no longer rejects classes that extend a sanctioned Effect base through a namespace import. `import * as Context from 'effect/Context'` followed by `class Foo extends Context.Service()('Foo')` was reported as banned: the rule dropped the namespace when resolving the base, so it compared `effect/Service` against its sanctioned list instead of `effect/Context.Service`. That spelling is the common one, so the rule was loudest exactly where the code was correct. Deep namespace imports such as `effect/Schema` resolve correctly too, and a member of a foreign module is still rejected. An ambient class declaration is no longer reported, whether it sits inside a `declare module` or stands at file scope as `declare class`. Neither emits anything at runtime, so none of the harms this rule prevents can occur there. ## 2.1.0 ### Minor Changes - The recommended set now enables three new rules at error. `internal-export-jsdoc` requires the `@internal` tag on every export that belongs in an internal module. `no-internal-jsdoc-outside` reports that tag on any other export. `tests-import-public-api` reports a test that reaches unpublished modules through a relative import. A name a public barrel re-exports is not internal — it lives next to the barrel, not under `internal/`. Spreading the recommended set is all it takes. ### Patch Changes - `no-domain-branching-density` now recognises a decision body passed as the second argument to `Workflow.make`, so such a body keeps the exemption it already had when it was the only argument. Bodies written in the two-argument form previously counted against the branching limit and could report. - The peer requirements for `effect` and for the Effect test-runner integration now accept any compatible `4.0.0-rc` release, instead of demanding one exact release candidate. Installing alongside a newer release candidate no longer reports an unmet peer dependency or resolves a second copy of `effect` into the dependency tree. ## 2.0.0 ### Major Changes - Removes `ban-classes`. Drop the entry from your config if you set it. The rule rejected every class declaration outside an allow-list of Effect constructors, which put `Context.Service` keys and hand-written adapters permanently in violation — reachable only by naming each one in a `whitelist` option. - Two rules are removed: `no-escaping-module-state` and `no-effect-service`. Both are gone from the shared base config, so a project extending it needs no change; a project naming either rule directly must delete that entry, and a project relying on either for enforcement should know neither was enforcing much. `no-escaping-module-state` refused a module-scope `Map`, `Set`, `Ref`, `Deferred`, `Queue` or semaphore. It read one file's syntax, so it decided the question only for the spelling it happened to see: the same primitive reached a consumer unreported through an object or array literal, a class static field, a `globalThis` assignment, a factory return, a destructured or computed member, or an immediately-invoked function. Its type-annotation arm was keyed on a closed list of type names, which meant every primitive absent from the list — the STM and concurrent families among them — passed, and every future one would need adding by hand. It also refused the shape its own message prescribed, reporting a module-private binding that was never exported at all. `no-effect-service` refused `Effect.Service`. Effect v4 exports no `Service` from the `Effect` module, so every use of it is already a type error naming the exact expression and offering the replacement — the rule restated a compiler diagnostic, and restated it only for a direct member access on a named import, missing a local alias, a computed key, a destructure, a namespace import, a cast, a re-export and a default import. Declare a service as `class X extends Context.Service()('id')`, adding a `make` option when the module owns its single construction. Build a coordination primitive inside the function that yields the surface and export the operations rather than the interior — `withLock(key, effect)`, `joinInFlight(key, effect)`, `read()` — so a caller depends on behaviour and swapping the interior or adding backpressure breaks nobody. ### Minor Changes - The plugin now recommends its own rules. `configs.recommended` lists every rule it ships at `error`, so a configuration can spread it instead of transcribing the rule names: ```ts import house from '@systemfsoftware/oxlint-plugin' export default { plugins: ['@systemfsoftware/oxlint-plugin'], rules: { ...house.configs.recommended.rules }, } ``` A hand-written list drifts silently — a rule added here never reaches a configuration that spelled its predecessors out, and a renamed rule is reported as unknown at most once. Every sibling plugin in this family already published this shape - The core regime keys on the `Workflow.make` boundary; the complement gains a complexity ceiling. - `workflow-match-exhaustive` no longer reads the filename: the gate is the make callee boundary (import binding + member `make` + argument containment, module-scope references followed, shadow-correct). Identical dispatch outside a make body produces no diagnostic. - New `make-body-purity`: references inside make bodies resolve only to parameters, const locals, and audited-pure imports; control flow is banned with the one first-statement converging guard; unclassifiable references report honestly as unresolvable rather than passing. Test files are exempt — fixtures exercise decisions without the production regime binding them. - New `no-domain-branching-density` in core: per-function McCabe CC outside make bodies, ceiling 17 — the lowest measured value the tree passes with zero waivers (max measured 17; the 15 functions over 10 are the recorded extraction backlog, not retrofit targets). - Both make-boundary rules fix classifier defects their first workspace run exposed: builtin globals with empty defs take the named-global triage, and `as const` type subtrees leave the value-reference walk. - cut over to effect v4 (4.0.0-rc.108): public surface derives from effect types; peers flip effect ^3→^4 ### Patch Changes - Array types are spelled one way. `Array` and `ReadonlyArray` in emitted declarations become `T[]` and `readonly T[]`, which the type checker cannot tell apart: no exported type changes, only how it is written. - New version is published through npm trusted publishing, so it carries a provenance attestation you can verify. - Rescope three rules that over-rejected legitimate files. Each was a measured false positive, not a theoretical one — all three fired on files written in the course of shipping `Workflow.make`: - `cell-suffix-required` rejected `src/Workflow.ts`, a PascalCase contract module whose basename names no cell because it _is_ the package's published contract. It now exempts that shape. - `no-manual-tag-member`, `no-direct-tag-access`, and `ban-data-taggederror` rejected `**/*.tst.ts` type-test fixtures, which must contain no runtime values and therefore cannot use `S.TaggedStruct`. The fix each rule suggested was impossible in the file it was suggesting it for. This is the cost the cell-taxonomy derivation already prices: a depth-0 rule cannot miss its target, but its predicate is a proxy for the property actually wanted, so it over-rejects — and that cost is meant to be priced per rule rather than waved away. Three rules, three over-rejections, all rescoped by narrowing the predicate; none was deleted and no config was weakened. ## 1.0.0 ### Major Changes - Removes `ban-classes`. Drop the entry from your config if you set it. The rule rejected every class declaration outside an allow-list of Effect constructors, which put `Context.Service` keys and hand-written adapters permanently in violation — reachable only by naming each one in a `whitelist` option. - Two rules are removed: `no-escaping-module-state` and `no-effect-service`. Both are gone from the shared base config, so a project extending it needs no change; a project naming either rule directly must delete that entry, and a project relying on either for enforcement should know neither was enforcing much. `no-escaping-module-state` refused a module-scope `Map`, `Set`, `Ref`, `Deferred`, `Queue` or semaphore. It read one file's syntax, so it decided the question only for the spelling it happened to see: the same primitive reached a consumer unreported through an object or array literal, a class static field, a `globalThis` assignment, a factory return, a destructured or computed member, or an immediately-invoked function. Its type-annotation arm was keyed on a closed list of type names, which meant every primitive absent from the list — the STM and concurrent families among them — passed, and every future one would need adding by hand. It also refused the shape its own message prescribed, reporting a module-private binding that was never exported at all. `no-effect-service` refused `Effect.Service`. Effect v4 exports no `Service` from the `Effect` module, so every use of it is already a type error naming the exact expression and offering the replacement — the rule restated a compiler diagnostic, and restated it only for a direct member access on a named import, missing a local alias, a computed key, a destructure, a namespace import, a cast, a re-export and a default import. Declare a service as `class X extends Context.Service()('id')`, adding a `make` option when the module owns its single construction. Build a coordination primitive inside the function that yields the surface and export the operations rather than the interior — `withLock(key, effect)`, `joinInFlight(key, effect)`, `read()` — so a caller depends on behaviour and swapping the interior or adding backpressure breaks nobody. ### Minor Changes - The plugin now recommends its own rules. `configs.recommended` lists every rule it ships at `error`, so a configuration can spread it instead of transcribing the rule names: ```ts import house from '@systemfsoftware/oxlint-plugin' export default { plugins: ['@systemfsoftware/oxlint-plugin'], rules: { ...house.configs.recommended.rules }, } ``` A hand-written list drifts silently — a rule added here never reaches a configuration that spelled its predecessors out, and a renamed rule is reported as unknown at most once. Every sibling plugin in this family already published this shape - The core regime keys on the `Workflow.make` boundary; the complement gains a complexity ceiling. - `workflow-match-exhaustive` no longer reads the filename: the gate is the make callee boundary (import binding + member `make` + argument containment, module-scope references followed, shadow-correct). Identical dispatch outside a make body produces no diagnostic. - New `make-body-purity`: references inside make bodies resolve only to parameters, const locals, and audited-pure imports; control flow is banned with the one first-statement converging guard; unclassifiable references report honestly as unresolvable rather than passing. Test files are exempt — fixtures exercise decisions without the production regime binding them. - New `no-domain-branching-density` in core: per-function McCabe CC outside make bodies, ceiling 17 — the lowest measured value the tree passes with zero waivers (max measured 17; the 15 functions over 10 are the recorded extraction backlog, not retrofit targets). - Both make-boundary rules fix classifier defects their first workspace run exposed: builtin globals with empty defs take the named-global triage, and `as const` type subtrees leave the value-reference walk. - cut over to effect v4 (4.0.0-rc.108): public surface derives from effect types; peers flip effect ^3→^4 ### Patch Changes - Array types are spelled one way. `Array` and `ReadonlyArray` in emitted declarations become `T[]` and `readonly T[]`, which the type checker cannot tell apart: no exported type changes, only how it is written. - New version is published through npm trusted publishing, so it carries a provenance attestation you can verify. - Rescope three rules that over-rejected legitimate files. Each was a measured false positive, not a theoretical one — all three fired on files written in the course of shipping `Workflow.make`: - `cell-suffix-required` rejected `src/Workflow.ts`, a PascalCase contract module whose basename names no cell because it _is_ the package's published contract. It now exempts that shape. - `no-manual-tag-member`, `no-direct-tag-access`, and `ban-data-taggederror` rejected `**/*.tst.ts` type-test fixtures, which must contain no runtime values and therefore cannot use `S.TaggedStruct`. The fix each rule suggested was impossible in the file it was suggesting it for. This is the cost the cell-taxonomy derivation already prices: a depth-0 rule cannot miss its target, but its predicate is a proxy for the property actually wanted, so it over-rejects — and that cost is meant to be priced per rule rather than waved away. Three rules, three over-rejections, all rescoped by narrowing the predicate; none was deleted and no config was weakened.