Kristoffer K.

> Looks like #4247, which bumped the cache key, was missed in this list. Bumping the cache key isn't a breaking change.

No, breaking changes can't be backported.

Hey, happy (late) new year! 😄 > there's a weird require It's something alright, fixed in https://github.com/facebook/jest/pull/13836. > WebAssembly is not defined, but AFAICT it should be part of TS...

> > I want to put more focus on the performance implications of not exposing this information. > > Why? Have you actually found this to be a major bottleneck...

> Speaking of rollup, I cannot make it understand require, and it also barfs on import thing = require('thing') even though babel is supposed to compile it away You might...

It's a bug in Node.js which was fixed in https://github.com/nodejs/node/pull/41221 and shipped in Node.js v14.19.2, v16.14, and v17.4.

One of the Yarn tests also noticed this change and was mentioned [on Slack](https://openjs-foundation.slack.com/archives/C053UCCP940/p1696025755249309), likely cause is https://github.com/nodejs/node/pull/49869.

> I think we should also do something about the builds in general, the AppVeyor stats doesn't seem to be reporting back to this repo as it should. Would be...

> if that bypasses yarn's bug. It's possible to mount the `node_modules` folder to a different drive using a different file system when using npm as well, shouldn't the check...

This issue is listed under [Milestone 2: Usability improvements](https://github.com/nodejs/loaders/tree/305e8da0fc22e43d7844034a865eba3b64c749a8#milestone-2-usability-improvements) but for parity with CommonJS it needs to be possible to register loaders at runtime since in CommonJS you can add...