Mingwei Samuel
Mingwei Samuel
Let's use `https://crates.io/crates/web-time` instead of `instant` (which is out of date) ```rust #[cfg(not(target_family = "wasm"))] pub use std::time::SystemTime; #[cfg(target_family = "wasm")] pub use web_time::SystemTime; ```
Oh! it seems the code in `instant` is actually just completely broken: https://github.com/sebcrozet/instant/issues/49 https://github.com/sebcrozet/instant/blob/master/src/wasm.rs#L201-L203 ```rust self.duration_since(SystemTime::now()) ``` Is backwards and would always be negative if time goes forward. Should be...
Yeah I knew to prefer web-time but I didn't realize how busted instant was until just now looking thru the issues ðŸ˜
I guess maybe your original bug cause may also be true?
> Do we want to put this reasoning up one level in Hydroflow+? Yes #1390
Some is in HF+ and some is flo properties in core (#1500)
Closing in favor of #1390 (HF+) and #1500 (HF core)
That's a good point
Could be implemented with an external channel (tokio channel) though maybe against the spirit of keeping things in hydroflow
Also interesting - if we split into a `sink_futures` and `source_stream` and the two operators are in the same stratum, then it may be weird if a future gets sent...