Sam Clegg

Basically `emscripten_exit_with_live_runtime` is a push without a pop, which means that runtime stays alive forever. Of course somebody could calls pop explicitly, but calling pop without a push should not...

> Primarily I want this to work for the proxying case. Unlike other Emscripten APIs, this one works via `Atomic.waitAsync`, and, unlike "regular" async APIs, atomics don't keep the event...

You can run `test/unistd/links.c` under wasmfs by running `wasmfs.test_unistd_links`, and this test does pass. However I suppose what doesn't work is symbolic links under OPFS? Is this expected behavior @tlively...

In terms of the webidl I think we can switch from `unsigned long` to `unsigned long long`. See #70. I'm not sure how/where we can make the argument type dependent...

Do you know if there is some way to represent polymorphic methods such as `get` and `set` that can take either an `unsigned long` or `bigint` depending on the type...

That is a scary warning. It does seems reasonable to allow both numbers and BigInt. For example, `.get()` and `.set()` on a table are logically equivalent to accessing elements in...

I guess the risk with going with JS numbers for now is that we would need to continue to support that going forward, even if we later decide to go...

@bvisness, thank you for the detailed analysis. I have been thinking along the same lines. i.e. That (2) is the way to go. Its unfortunate since it will break existing...

It should be possible do with a pair of i64 maybe? Not sure its worth the effort though, especially since autodebug is a very niche option.

Closing this issue since this proposal is done and the repo is being archived. Any further discussion can take place in the spec or design repos.