Is there interest in a Volt-native test runner?

[MuNeNiCK]

Hi, I wanted to ask whether there are any plans or thoughts around testing support for Volt projects.

I am currently trying to migrate an Elixir + Vite application to Volt. Volt seems like it can cover the build side of that workflow well, but I was not sure what the intended direction is for frontend tests that are currently run with Vitest.

To explore this, I put together a small experimental repository called Voltest:

https://github.com/MuNeNiCK/voltest

The idea is to run frontend test files through Volt itself: compile test modules with Volt, resolve imports using Volt plugins and aliases, and execute the compiled bundle from a Mix task.

The repository currently includes a small initial test API:

• describe, test / it, skip, only, each
• beforeAll, beforeEach, afterEach, afterAll
• common expect matchers
• some vi helpers, such as vi.fn, vi.spyOn, and mock reset / restore helpers
• top-level vi.mock support through a Volt transform
• a few browser-like primitives, such as Headers, Request, Response, location, events, and storage

This is still experimental, and I mainly wanted to make the idea concrete enough to discuss.

If Volt already has a direction for testing, I would be interested in learning more about it.

If this direction seems useful, would it make sense for something like this to live under the elixir-volt organization?

I would be happy to transfer the repository if that helps, or adapt it in whatever way best fits Volt. The name, implementation, and scope are all flexible.

[dannote]

Thanks for exploring this — the direction is definitely interesting.

I’ve been thinking about testing support for Volt, but haven’t decided yet what the right shape should be. There are a couple of possible layers here: a JS-compatible API/runtime for easier migration from tools like Vitest, and an ExUnit integration that feels natural in Elixir projects.

My current instinct is that ExUnit integration should probably exist, since it’s the idiomatic entry point for Elixir users and would make Volt tests fit naturally into mix test. At the same time, a JS API may still be valuable for compatibility and migration, especially for projects already using Vitest-style tests.

So I’m not sure yet whether the right answer is ExUnit, a JS compatibility API, or both — but I suspect both may make sense if the layers are kept clean. The Voltest experiment looks like a useful concrete starting point for that discussion. I’ll look into this soon.