chore: migrate to vite-plus unified toolchain

[afonsojramos]

Summary

• Replace @biomejs/biome with Vite+ (vite-plus@0.1.20), which bundles oxlint, oxfmt, Vitest, and the dev/build pipeline behind a single vp CLI.
• Consolidate linter, formatter, and pre-commit configuration into the lint, fmt, and staged blocks of vite.config.ts — .oxlintrc.json, .oxfmtrc.json, and the lint-staged block in package.json are gone.
• Switch package scripts to vp dev / vp build / vp test / vp check, swap husky's pre-commit to vp staged, and point VS Code's oxc extension at vite.config.ts.
• Convert all biome-ignore directives to oxlint-disable, fix the lint findings oxlint surfaced (constant-binary-expression, unsafe optional chaining, unused param, useNavigate stability), regenerate one snapshot affected by oxfmt's JSX whitespace, and apply oxfmt formatting across the codebase.

Test plan

• pnpm check (vp lint + vp fmt: clean across 339 lint files / 373 fmt files)
• pnpm tsc --noEmit
• pnpm test (1031/1031 passing)
• pnpm build

[setchy]

No vite+?

[setchy]

Does vite-plus mean we should drop vite and vitest as direct devdeps?

[afonsojramos]

In principle yes — vite-plus aliases vite (@voidzero-dev/vite-plus-core) and vitest (@voidzero-dev/vite-plus-test) and the docs recommend dropping the standalone deps after rewriting imports to vite-plus / vite-plus/test.

I tried it and the rewrite forces tsconfig moduleResolution to switch from "Node" to "Bundler" so TS can resolve the subpath exports. Bundler resolution then exposes pre-existing latent issues that Node resolution masked:

• src/renderer/types.ts imports from @primer/react/dist/Button/types (deep import not in @primer/react's exports map)
• src/renderer/utils/ui/theme.ts imports from @primer/react/dist/ThemeProvider (same)
• @testing-library/jest-dom matchers (toBeInTheDocument, toHaveTextContent) stop being picked up via vitest/globals types and need to be wired through vite-plus/test/globals or via setupFiles types

Each is fixable but the diff balloons well beyond the toolchain swap. I reverted those changes and would rather chase the cleanup in a follow-up.

[afonsojramos]

Took another pass. Two of the three issues are fixable cheaply, but dropping vite or vitest themselves still hits ecosystem walls:

Fixed in cfcfa5a (regardless of devdep cleanup):
- src/renderer/types.ts and utils/ui/theme.ts now inline VariantType and ColorModeWithAuto so we no longer reach into @primer/react/dist/*

Ended up reverting the above. Does not feel like it is worth it at the moment.

Conclusion, could not drop vite from devdeps: vite-plugin-electron does require('vite') at the project root. With vite removed, pnpm only places vite-plus-core's vite alias inside vite-plus-core itself, so the plugin's resolve fails. Re-adding vite restores the build.

Could not drop vitest from devdeps either: vitest's matchers are augmented via @testing-library/jest-dom/vitest, which does declare module 'vitest'. Once vitest is gone the global expect is typed as @voidzero-dev/vite-plus-test's Assertion (declared inside dist/chunks via a relative module spec that isn't reachable from outside the package's exports map). I tried augmenting via the package, the dist subpath, and tsconfig paths; none of them merge with the inner declaration. Keeping vitest as a devdep keeps the existing augmentation working.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
89.4% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud