| 07:50 | <annevk> | Kris Kowal: I guess I'd also expect them to be mutually exclusive (either through an error or ignoring), but putting a request for metadata in the same namespace as a request for data feels weird |
| 14:54 | <ljharb> | import.resolve with consistent cross-platform behavior would be awesome |
| 14:54 | <ljharb> | there's currently some contention about sync vs async in browsers vs node iirc |
| 14:55 | <nicolo-ribaudo> | Node.js recently changed it to be synchronous to align with browsers |
| 15:00 | <ljharb> | right - which is "fine" in the sense that i prefer sync, but it's not fine in the sense that node needs it to be async for various reasons, and also in the sense that node doing something bad for node to align with browsers is hugely harmful for node imo (whereas, if we work out the issues in the spec then they're all aligning with the spec, and that's good) |
| 15:38 | <littledan> | were these reasons around the semantics of soft links? |
| 15:38 | <littledan> | being the same cache key (vs different for HTTP redirects) |
| 15:41 | <annevk> | They might have to do I/O to resolve, I think. I think they might have also followed symlinks? Very different specifier behavior. |
| 20:29 | <ljharb> | i don't recall the specifics |
| 20:30 | <ljharb> | for all i know the requirement was ideological instead of technical |
| 21:19 | <littledan> | for all i know the requirement was ideological instead of technical |
| 21:20 | <ljharb> | indeed |