chore(deps): upgrade vite-plus to 0.2.6

[shulaoda]

What

Bumps the build toolchain from 0.2.2 to 0.2.6 via the workspace catalog:

• vite-plus: 0.2.2 → 0.2.6
• vite (aliased to @voidzero-dev/vite-plus-core): 0.2.2 → 0.2.6

These two move together by requirement — vite-plus@0.2.6 depends on @voidzero-dev/vite-plus-core@0.2.6. Also includes the small non-shipping changes needed to stay green under the new toolchain (see Toolchain fallout), plus a merge with main to regenerate the lockfile for a newly-added example.

Why

Keeps the vinext build/test toolchain current and picks up the fixes shipped across 0.2.3–0.2.6. vite-plus is a root devDependency (build tooling only), so nothing here changes any published package's runtime.

How

Edited the two catalog entries in pnpm-workspace.yaml:

-  vite: npm:@voidzero-dev/vite-plus-core@0.2.2
+  vite: npm:@voidzero-dev/vite-plus-core@0.2.6
-  vite-plus: 0.2.2
+  vite-plus: 0.2.6

Then regenerated pnpm-lock.yaml. Re-resolving the core package naturally moves its bundled toolchain subtree — all transitive, all within the vite-plus dependency graph:

• @voidzero-dev/vite-plus-core + platform bindings → 0.2.6
• oxfmt → 0.60.0, oxlint → 1.75.0, oxlint-tsgolint → 7.0.2001, @oxc-project/{runtime,types} → 0.141.0
• yuku-parser / yuku-codegen → 0.5.48
• lightningcss → 1.33.0
• @vitest/browser* → 4.1.10

The catalog-pinned vitest@4.1.9 and @vitest/coverage-istanbul@4.1.9 are unchanged, and no dependency outside the vite-plus subtree drifted.

Toolchain fallout

The 0.2.6 toolchain (oxfmt 0.60.0, oxlint 1.75.0, rolldown/oxc) surfaced three CI adaptations. All are tooling/test-only — none change a published runtime:

1. Formatting — oxfmt 0.60.0 reformats README.md, apps/web/next.config.ts and tests/nextjs-compat/TRACKING.md (collapses empty-object-with-comment onto one line; unpads markdown tables). Ran vp check --fix.
2. Lint — oxlint 1.75.0 now flags dynamic import("node:path") under the existing no-restricted-imports rule (previously only static imports). Disabled inline in loadStaticPrerender with a reason — the resolved path feeds a dynamic import(), so pathslash's forward-slash canonicalization buys nothing there.
3. font-google-build unit test — rolldown now code-splits the server bundle, so the font markers (Geist, selfHostedCSS, …) moved out of index.js into _next/static/* chunks. Verified they still exist (relocated, not a regression) and updated the test to scan the whole server output rather than index.js alone.

Lockfile

Merged latest main and regenerated pnpm-lock.yaml: main added examples/pages-router-complex, whose vite: catalog: dep now resolves to 0.2.6. Only that importer's vite / vite-plus / @cloudflare/vite-plugin entries changed.

Note

While investigating the failing shards, one app-router e2e (metadata-icons.spec.ts → "rapid icon replacement") can surface BodyStreamBuffer was aborted: a pre-existing, timing-dependent race where a fast follow-up navigation aborts a still-streaming RSC fetch and React reports the un-consumed metadata Flight chunk's AbortError globally. The new toolchain shifts navigation timing so it can flake on CI. Left out of scope here to keep this a pure toolchain bump; if it recurs it can be addressed separately (suppress the benign abort in the runtime, or ignore it in the test harness).

Changeset

devDependency + test/format/lint only — no published runtime change, so no changeset / version bump.

[pkg-pr-new]

Open in StackBlitz

npm i https://pkg.pr.new/@vinext/cloudflare@2694

npm i https://pkg.pr.new/create-vinext-app@2694

npm i https://pkg.pr.new/@vinext/types@2694

npm i https://pkg.pr.new/vinext@2694

commit: 02b1c22

[github-actions]

Performance benchmarks

Compared 02b1c22 against base 05eee9f using alternating same-runner rounds. Next.js was unchanged and skipped.

1 improved · 2 regressed · 3 within ±1.5%

Scenario	Framework	Baseline	Current	Change
Client bundle size (gzip)	vinext	132.4 KB	132.3 KB	⚫ -0.1%
Client entry size (gzip)	vinext	119.7 KB	119.6 KB	⚫ -0.1%
Dev server cold start	vinext	2.78 s	2.77 s	⚫ -0.4%
Production build time	vinext	2.92 s	2.87 s	🟢 -1.6%
RSC entry closure size (gzip)	vinext	103.3 KB	108.1 KB	🔴 +4.7%
Server bundle size (gzip)	vinext	177.8 KB	184.7 KB	🔴 +3.9%

View detailed results and traces

_{🟢 improvement · 🔴 regression · ⚫ change below 1.5% · paired base/head}