Flashes Home Search Notifications Sign in Post

VoidZero

@voidzero.dev

3308 Followers 24 Following

The JavaScript Tooling Company

Bar chart showing per-version downloads for 5 major versions of the vite package, from 24 Jun to 30 Jun, 2026 from the v4.x version to the v8.x version. The most downloaded version is v8.x with 47.3m downloads. v7.x version was downloaded 43.7m times, v6.x version was downloaded 19.9m times, v5.x version was downloaded 28.2m times, v4.x version was downloaded 2.5m times.

VoidZero + Cloudflare

crates.io screenshot showing the Rolldown crate

Demoing the new org template feature

Showcasing React Compiler rules with Oxlint

Alex showing the Oxfmt jsdoc formatting

Explainer of Vite 8's new React plugin major

Showcase of Bun support

TL;DR of the Void announcement

A screenshot of the docs showcasing the agent reporter feature

Screenshot of the setup.viteplus.dev page showing a download button and info about which (Windows) distributions are supported.

Demo of tsdown SEA support

A screenshot of the GitHub repository of vite-plus showing the "Private" label.

Video showing the "find slow test" feature and filter in Vitest UI

Info video about the Oxfmt beta

A screenshot of VIte 8 download numbers showing 2.610.727 downloads in the last 7 days while it was published 7 days ago

Grant Gustin posing next to a grave meme: the gravestone is labeled "JavaScript Fatigue" and Grant Gustin, kneeling beside the grave with a peace sign, is labeled "Vite+"

Git changes where .lintstagedrc, .prettierrc, eslint.config.js, vitest.config.ts, and the husky pre-commit hook are deleted, while package.json, pnpm-lock.yaml, and vite.config.ts are modified. A pixelated file is added.

"A dark-themed code editor showing three JavaScript import statements. The first imports defineConfig from a partially pixelated module name starting with "vi", the second imports a pixelated export name starting with "v" from a pixelated module also starting with "v", and the third reads import { evan } from "@vuejs/amsterdam".

Screenshot of VoidZero supporting npmx with $500 USD / month

Infographic showing Rolldown's entriesAware code splitting feature in 3 steps. Step 1 shows three entry points (Dashboard using chart.js, Admin using marked, Landing using gsap - all sharing es-toolkit). Step 2 shows the problem: without entriesAware, all entries load a single 305KB vendor.js containing all libraries. Step 3 shows the solution: with entriesAware enabled, chunks are split by usage - a shared 8KB chunk for es-toolkit plus entry-specific chunks (195KB for dashboard, 40KB for admin, 63KB for landing), reducing average page load to 107KB and eliminating unused library downloads. Code snippet shows the configuration: entriesAware: true and entriesAwareMergeThreshold: 20000.

Video showcasing the new Vitest detect async leak flag

"Vitest 4.1 introduces Test Tags. Left side shows a vitest.config.js code snippet defining three tags: 'frontend' with a description, 'db' with a 60-second timeout, and 'flaky' with conditional retry logic for CI and priority setting. Right side shows CLI filtering examples: vitest --tags-filter="frontend && !flaky", vitest --tags-filter="(frontend || db) && !flaky", and vitest --tags-filter="frontend and flaky". VoidZero and Vitest logos at bottom corners on dark background."

Rolldown - Lazy Barrel Optimization

Infographic titled "Empty tests marked as todo." Shows a Vitest test file importing describe and it from 'vitest', with a test suite containing an it('should to something in the future') call that has no callback function. Below is Vitest's terminal output from running the test, showing "content.test.ts (1 test | 1 skipped)" with the test marked as "1 skipped" and "1 todo (1)". Yellow arrows connect the empty test definition to the corresponding output. Explanatory text at the bottom reads: "A test without a function is automatically marked as todo and will be skipped in Vitest." VoidZero and Vitest logos at bottom corners.

Infographic titled "Vitest's expect.poll" with three bullet points: built-in timeout and retry logic, cleaner than wrapping assertions in vi.waitFor, and perfect for job queues & eventual consistency. Below is a code example showing a test that starts a background export job, then uses expect.poll to repeatedly check getJobStatus(job.id) until it equals 'completed'. VoidZero and Vitest logos at bottom corners.

Graphic titled "Rolldown exposes parse function." Left side shows a JavaScript code example: importing parseSync from "rolldown/experimental", parsing the string "const a: number = 1;" as TypeScript, and logging the resulting AST, an object with type "Program" containing a VariableDeclaration node with properties like kind, declarations, start, and end positions.

Right side lists key points: Rolldown provides parse/parseSync, it's re-exported from Oxc for high performance, also exposed in Vite 8, usable in Rolldown/Vite plugins, and reduces dependencies (no separate Oxc install needed). VoidZero and Rolldown logos at bottom corners.

Collage of all available mobile backgrounds.

Chart of weekly Vitest downloads, showing a peak of 20.18 million downloads at the beginning of December 2025.

$Rolldown REPL with rolldown.config.ts ```ts import { defineConfig } from "rolldown"; export default defineConfig({ output: { minify: true, postBanner: '// Appended to each file', footer: '// I\'ll be minified and GONE', postFooter: '// Prepended to each file' } }); ``` Bundled `index.js`: ```js // Appended to each file console.log(`abc`); // Prepended to each file ```$

A JAEGER UI chart showing a run of vitest with data collected through the OpenTelemetry integration.

A screenshot of the terminal when building your Vite app. It says "[PLUGIN_TIMINGS] Warning: Your build spent significant time in plugins. Here is a breakdown:".
Then a list of plugins and percentages. The list includes "vite:terser", which could be fixed by using Oxc instead of terser as minifier.

Title: Import and transform time breakdown

Three screenshots from the Vitest UI, the terminal and the VS Code Extension showing the time it takes to import a module.

Vitest’s Experimental FS Cache
Enable via CLI or config: `pnpm vitest --experimental.fsModuleCache`

First run does not change timings. There is no cache yet!
An image with statistics about the test run with 3088 tests passed and 6.17s testing duration, transform taking 5.32s, setup 461ms and collect 18.50s.

Subsequent runs use the file system cache, reducing transform times
Another statistics image. The subsequent run now takes only 5.14s, with transform time down to 673ms, setup to 341ms and collection time down to 13.20s

Title: Improved require(esm) support
Subtext: chunk.mjs defines the return value when used via require by setting it via export `as “module.exports"`

Image showing that index.js requires chunk.mjs and logs result:
const User = require('./chunk.mjs');
console.log('User:', User);

The chunk exports a user class by default, but also via "export { User as 'module.exports' }".
---
Screenshot of the a previous Rolldown version showing the console.log that does not log the class but an object with default and 'mode.exports' key.
Subtitle: OLD: Use literal exports, no transformation

Screenshot of the latest Rolldown version correctly logging that User is a class.
Subtitle: NEW: Detection & correct runtime result

Three bullet points at the end:
* More predictable behavior
* Aligns with Node.js
* Only relevant for require(esm)

Three column layout.
First column: "Input" shows an index.ts with the following content

```js
const url = new URL('./styles.css', import.meta.url)
console.log(url)
```

It also shows a styles.css but its content is not important.
Subtitle: Code to get a reliable URL to a local asset Runs everywhere, right?

Second column: "Before"

index.js:
```js
//#region index.ts
const url = new URL("./styles.css", import.meta.url);
console.log(url);
//#endregion
```

Shows a mismatch between the "./styles.css" url and the bundled "./styles-DvhEFzjT.css" tab that is also included in the column.
Subtitle: MISMATCH - After bundling, URL does not contain the updated asset name with the file hash.

Last column: After

Shows a rolldown.config.ts that enables `experimental.resolveNewUrlToAsset`.

Then, shows an index.js that now has the right reference to the bundled css file, "assets/styles-CjdrdY7X.css":
```js
const url = new URL("assets/styles-CjdrdY7X.css", import.meta.url);
console.log(url);
```

Subtitle: With the experimental option, relative assets are now referred to correctly ✅

Title: Match against Standard Schema in Vitest

Code

```js
import { it, expect } from "vitest";
import * as v from "valibot";
import * as z from "zod";

const schema = v.object({
name: v.string(),
age: v.number(),
email: v.pipe(v.string(), v.email()),
});

it("should match the schema", () => {
const user = {
name: "Jane Doe",
age: 42,
email: "youshoulduse@vitest.dev",
};

// Validate the entire object
expect(user).toEqual(expect.schemaMatching(schema));
// Also works for validating partial data
expect(user).toEqual({
name: 'Jane Doe',
age: 42,
email: expect.schemaMatching(z.string()),
});
});
```

Result when running Vitest: Test passing.

Input: index.ts with two lines:
const a = new Int32Array()
const b = new Uint8Array()

Subtitle: Any kind of built-in typed array constructor that is unused (dead code)

---

Result Before - index.js:
new Int32Array();
new Uint8Array();

Subtitle: No variables anymore, but unused constructor call still exists

---

Result after: empty index.js
Subtitle: Code is fully removed ✔

Legal

Privacy Policy Terms and Conditions

Contact

Follow

Bluesky Instagram Threads

Flashes for Bluesky Get it on the App Store