Yeah, I have prepared our company software for migration back to node.

I would like to read the promised Jarred's blog post (if it ever comes out) before pulling the plug though.

I’m amazed folks ever migrated to it at a workplace level TBH. It’s a VC funded runtime, something was always going to happen to it. Node is boring but its governance and ownership is clear.

Well, it was (is?) significantly easier to use, faster, the team behind it is fantastic, and generally had enough compatibility to run the projects I needed it to.

When it’s enough of a drop-in replacement, that’s more than good enough. As long as we’re not adding a hundred bun-specific things, it’s not terribly difficult to back out of, either. Kind of a no-brainer.

Faster I get. But easier to use?

There is no comparison when running a TypeScript project with Bun vs with Node. Or when mixing ESM and CJS. Or when setting up unit testing.

Node and its ecosystem was considerably behind in multiple regards.

OP said:

> it’s not terribly difficult to back out of, either

Things like ESM/CJS interop are great but backing out of an ecosystem that uses it is absolutely not trivial.

Either it’s a trivial replacement you can drop in with no changes or it’s a notable change you can’t easily switch out. It’s not both.

Yeah. Take a look at the docs. So much niceties

...and now we also have Nub :

https://nubjs.com/blog/introducing-nub

Take a look at Deno. It's pretty great and now has exceptional node compatibility

> Take a look at Deno.

It's still heavily AI driven. Maybe less than Bun.

And they've gone in the route of just taking Node native libraries and what not just because they gave up on working out the compatibility. It's a bit of tacked on mess now.

Where are you seeing that it's heavily AI driven?

---

Edit: the contribution guidelines allow AI-assisted patches with disclosure. Also, there are a bunch of recent commits co-authored by bots