Stephen Belanger

Well, the first would fail when trying to execute a function that doesn't exist, but the second should work.

As the current maintainer of async_hooks, I will say there's basically zero chance of such a change being accepted. We're currently on a path to _less_ internals exposed, not _more_,...

Is there a reason we can't follow the Deno model but have permissive by default _unless_ there are --allow-* flags set? It _should_ mean no ecosystem breakage. Also gives us...

Well, you could just leave it on the user to give all the necessary permissions, including what Node.js itself needs. I would argue that's probably the safer approach anyway. It's...

Well, can we not alter our module resolution to halt when it reaches a level it doesn't have access to? It should be starting at the module path and walking...

Loaders doesn't help us for CJS, which is the vast majority of code out there. I feel we should also bring up the mechanism by which this additional phase is...

APM doesn't really care that cloud function platforms have their own bootstrap stuff in there. We want to be to run when the platform starts the Node.js process because otherwise...

Interesting. I'll have to remember to find some time to investigate that. 🤔

Unless the startup phase loop can be resolved and a new one started with the application entrypoint, there's not likely even a path to a solution for async_hooks.

Also, AsyncLocalStorage, which is built on top of async_hooks, is marked as stable. We can't just break compatibility with async_hooks because it would break AsyncLocalStorage too. There is a rather...