A non-`Error` value thrown from a test environment's `setup()` is dropped from the "Failed to start worker" report (regression from 3.x)

[brice-stacks]

Describe the bug

When a custom test environment's setup() rejects, the pool wraps the failure as
[vitest-pool]: Failed to start <pool> worker for test files … and attaches the
original value via { cause }. However, the reporter only prints the cause when it
is an Error instance:

• throw new Error("…") → report includes Caused by: Error: … ✅
• throw "…" / throw { … } / a wasm-bindgen JsValue → the cause is silently dropped; the user sees only the generic, location-free pool message ❌

In Vitest 3.x the same non-Error throw was surfaced directly (Unknown Error: <value>),
so this is a regression. It appears to have come in with the worker-start error handling
added in #9337 / v4.0.17 (which correctly fixed the indefinite hang from #9271, but the
new wrapper only renders Error-typed causes).

This bites real-world environments that initialize native/wasm resources in setup()
and reject with a non-Error value. For example @stacks/clarinet-sdk's
vitest-environment-clarinet calls a wasm-bindgen initSimnet() in setup(); when a
contract fails to compile, the wasm rejects with a JsValue, and on v4 the actual
diagnostic is completely lost — you only see Failed to start forks worker.

Reproduction

https://stackblitz.com/edit/vitest-dev-vitest-ldzxobzz?file=throwing-env.js

package.json

{
  "name": "vitest-env-setup-error-swallowed",
  "version": "1.0.0",
  "type": "module",
  "private": true,
  "scripts": {
    "test": "vitest run"
  },
  "devDependencies": {
    "vitest": "^4.1.8"
  }
}

throwing-env.js

// Minimal custom test environment whose async setup() rejects with a NON-Error value.
// Real-world analogue: wasm-bindgen / native addons reject with a plain JsValue
// (e.g. @stacks/clarinet-sdk's simnet init), not an `Error` instance.
export default {
  name: "throwing-env",
  transformMode: "ssr",
  async setup() {
    // A proper `throw new Error("...")` IS surfaced by vitest (as `Caused by:`).
    // A non-Error value is silently dropped — only the generic pool error remains.
    throw "the real reason setup failed (a non-Error value)";
  },
};

vitest.config.js

import { defineConfig } from "vitest/config";

export default defineConfig({
  test: {
    environment: "./throwing-env.js",
    // pool: "forks" is the default; the message mentions "forks worker".
  },
});

test/example.test.js

import { test, expect } from "vitest"; test("noop", () => { expect(1).toBe(1); });

Run npm install and npm test.

(Swap the throw for throw new Error("…") to see the cause correctly printed — that
contrast shows the wrapping/cause machinery is present; only non-Error rendering is
missing.)

System Info

System:
    OS: macOS 26.5.1
    CPU: (14) arm64 Apple M4 Pro
    Memory: 120.22 MB / 48.00 GB
    Shell: 5.9 - /bin/zsh
  Binaries:
    Node: 25.1.0 - /Users/brice/.nvm/versions/node/v25.1.0/bin/node
    npm: 11.6.2 - /Users/brice/.nvm/versions/node/v25.1.0/bin/npm
  Browsers:
    Chrome: 149.0.7827.55
    Safari: 26.5
  npmPackages:
    vitest: ^4.1.8 => 4.1.8

Used Package Manager

npm

Validations

• Follow our Code of Conduct
• Read the Contributing Guidelines.
• Read the docs.
• Check that there isn't already an issue that reports the same bug to avoid creating a duplicate.
• Check that this is a concrete bug. For Q&A open a GitHub Discussion or join our Discord Chat Server.
• The provided reproduction is a minimal reproducible example of the bug.