Overview - Router

routerWorkspace

CI Pipeline Executions
- Filtered
Runs
Demo

Insights

Compare tasks
Analytics

Overview

router

Loading workspace stats

Loading workspace insights...

Statistics interval

7 days30 days

Latest CI Pipeline Executions

Filter

Fuzzy

Sort by

© 2026 - Nx Cloud

Terms of Service Privacy Policy Changelog Status Docs Contact Nx Cloud Pricing Company @NxDevTools

routerWorkspace

Average execution time

9m 55s

192 CI pipeline executions

Computation time saved

41d 57m

Cache hit rate

50%

Time to Green

1h 22m

Was 2h 27m, improved by 44%

Succeeded
8504
1cc81971 refactor: standardize Yuku compiler helpers and terminology
by Sheraff
Succeeded
8504
37111c76 docs(changeset): state the createServerFn placement build error and the route factory warning Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
by Sheraff
Succeeded
8504
ca9f8dc1 chore: satisfy lint and type checks of the compiler regression fixes Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
by Sheraff
Succeeded
8504
4e3045b8 docs(router-plugin): describe split-module export imports without PR history Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
by Sheraff
Succeeded
8504
63503002 test(router-plugin): drop the duplicate type and value re-export case Rejecting `export type { X } from` plus `export { X } from` is the parser following TypeScript, not code-splitter behaviour, and the input already fails type-checking. Keep only the valid case of a type and a value declared with the same name and exported together. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
by Sheraff
Canceled
8504
e595df04 test(router-plugin): reject duplicate type and value re-exports like TypeScript `export type { Post } from './a'` plus `export { Post } from './b'` is a duplicate export for TypeScript ("Duplicate identifier 'Post'"). Babel accepted it, but the compiler now follows TypeScript and fails with a clear error naming the binding, so the test expects that rejection. A type and a value declared with the same name and exported together stay valid. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
by Sheraff
Canceled
8504
d8961293 chore: upgrade Yuku to 0.18.0 0.18.0 fixes the upstream issues the compiler tests were waiting on: deep expression trees decode without overflowing (#228), `import x = ns.T` resolves `ns` as a namespace (#229), codegen keeps decorators before `export` (#230), JSX text is entity-decoded (#231), and duplicate exports follow TypeScript (#232). It was published today, so allow its packages in minimumReleaseAgeExclude until they pass the one-day release age. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
by Sheraff
Failed
8504
4aede324 Merge branch 'main' into codex/yuku-evaluation Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Succeeded
8504
09fe8528 fix(router-utils): remove unused `var` declarators nested in statements A `var` declared inside an `if`, `try`, `switch` case, labeled block, `for` head or a block inside a function had no removable declaration, so it survived in every output that no longer used it: server-only initializers stayed in Start client and SSR callers, and the code splitter left component-only initializers in the reference module (running them twice) and copied loader-only ones into component chunks. Main removed them. The declaration index now separates the two roles the earlier block-scoped `var` fix had merged. Every declarator is a removable declaration, except a for-in or for-of head that the loop assigns. Only a declaration that is a statement of its binding's own scope is the binding's declaration to share or import; this also excludes a `var` in a statement slot such as `if (x) var y = z`, which the code splitter used to share. Removing the last declarator of such a slot leaves an empty block, and of a `for` head an empty init, so the enclosing statement stays valid. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Succeeded
8504
39abc15f test: cover dead-code and RSC transform cases that main miscompiles Ported from babel-dead-code-elimination, Turbopack's tree-shaker fixtures and @vitejs/plugin-rsc transform tests. These pass with the Yuku compiler and fail on main: empty destructuring patterns keep their initializer; component- and handler-only overloads, enums with namespaces, classes with static instances and redeclared vars leave the route module or client; assignments before a `var` or function declaration render correctly; a handler-only import reused as an enum member, overload parameter or parameter property is dropped from the client; `(x as T)++` and `(x satisfies T)++` print with parentheses. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Succeeded
8504
967638b1 fix(router-utils): keep solid-refresh pragmas leading compiler output `@refresh reload` and `@refresh skip` configure solid-refresh for the whole module, like React Refresh's `@refresh reset`, but the shared file-pragma rule only kept the latter. Split chunks and reference modules lost them, so Solid routes stopped reloading or skipping as the file asked. Also add React Refresh cases that main miscompiles (ported from the React Refresh Babel plugin tests): `as`/`satisfies`/non-null inline components are hoisted and registered, inline `memo`/`forwardRef` split components compile, `export { page }` components keep a valid reference module, and a function shared by two component options is declared in the chunk. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Canceled
8504
6ecbfdd4 test(start-plugin-core): cover directive-compiler cases that main miscompiles Ported from SolidStart's "use server" compiler and Waku's server-function tests. These pass with the Yuku compiler and fail on main: a nested server fn named like a top-level one no longer creates a phantom registration and an invalid provider; client-only and isomorphic fns inside a handler are compiled out of the provider; enums and namespaces read by Hydrate children reach the chunk; parameters and locals named like Start factories are not compiled as factories. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Canceled
8504
36ed1262 refactor(router-utils): share one file-pragma rule between the compilers The code splitter and the Start compiler each kept the source's top-level pragma comments leading their output, with two copies of the same helper. Move it to router-utils as `keepFilePragmas` and use it from the splitter's module generation, the Start compiler and the Hydrate chunk loader. The Start compiler now also keeps `@refresh reset`, as the splitter did. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Succeeded
8504
271eb00c test: cover React Compiler fixture shapes that main miscompiles Ported from React Compiler fixtures wrapped into route files, Start modules and Hydrate children. These pass with the Yuku compiler and fail on main: split components referenced by other exports, `export default function` rendered by another split component, lowercase components shared with the loader under React HMR, Hydrate children using `<_Bar>`, member tags, default-exported components or a module `let` they write, and server-only functions shadowing their own name. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Succeeded
codex/yuku-evaluation
dfc1cdd7 test: cover Hydrate and code-splitting cases that main miscompiles Ported from Qwik optimizer tests. These pass with the Yuku compiler and fail on main: Hydrate children using member JSX tags (`<ui.Home/>`) or names starting with `_`/`$` capture their object; optional members and method keys are not captured as locals; default-exported functions, enums and namespaces used by the children reach the chunk; `useId?.()` is rejected; a split chunk keeps a namespace's body. validate-module.ts is an identical copy of the helper added by #8629. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Canceled
8504
16dac799 test(start-plugin-core): cover server-fn cases that main miscompiles Ported from Next.js server-actions fixtures. These pass with the Yuku compiler and fail on main: - server fns not assigned to a variable (destructuring declarators, static class properties, JSX attributes, assignments, conditionals, a default export next to a declared fn) fail the build instead of shipping the handler and its server-only imports to the client; - comments around a handler do not leak into client and SSR caller output; - classes and enums used only by a handler are removed from the client, together with their server-only imports. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Succeeded
8504
c025def0 fix(start-plugin-core): count every JS line terminator in compiler diagnostics Compiler warnings and code frames split source only on \n, so files with lone \r, U+2028 or U+2029 line terminators reported wrong lines and framed the wrong source. Count line terminators the way JavaScript, editors and the parser do. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Succeeded
8504
70188c0b test(start-plugin-core): fail the build for server functions not assigned to a variable A createServerFn that is a default export, an object property, a function's return value or a call argument cannot be extracted to the provider module. Leaving it untransformed would ship the server handler to the client, so the compiler throws in every environment. Nothing covered this: a fast path that skips these calls passed the whole suite. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Succeeded
8504
91bdae04 fix(start-plugin-core): accept direct-call kinds that were not text-detected Resolved candidates were filtered by the text-detected kinds, so an aliased env-only factory (e.g. a re-exported `createClientOnlyFn as clientOnly`) was left untransformed when the file only mentioned another factory such as `createServerOnlyFn`. Filter resolved candidates by the environment's lookup kinds instead; text detection still decides which calls are candidates. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff
Succeeded
8504
6d3bb1df chore(router-utils): silence no-unnecessary-condition on the CJS import.meta fallback Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
by Sheraff