Post

⚡️ Yes! Vitest is generally faster than Jest -- but is feedback speed the only factor that matters? If you are still on the fence about switching to Vitest, here are 15+ reasons that will convince you! 📝 cookbook.marmicode.io/angular/why-... #javascript #angular #testing

Great post! Vitest does indeed allow users to choose the isolation mode and parallelization method (process, threads) based on their needs. The VM isolation on Vitest has same issues as Jest; node:vm leaks memory and we need to restart forks/threads when a certain limit is reached to prevent OOM.

Choosing between forks or threads is also great, as there are many packages that cause segfaults when run on 'node:worker_threads'. Typically these packages are built using Node-API with C++/Rust/Go. We've listed such packages in github.com/vitest-dev/v.... None of those packages crash with forks.

It's also worth to note that comparing non-isolation mode against Jest isn't really fair, and we've seen even ~10x faster execution times on Vitest without isolation: github.com/vitest-dev/v... Though I think there is a community-driven Jest runner that provides non-isolation mode there too.

 Absolutely! You are referring to this I guess: github.com/nicolo-ribau... I did my best to stay as fair as possible by including benchmarks of different isolation modes + cold start vs. warm.