Temporal support (TC39 stage 3 proposal)

[notramo]

What is the problem this feature would solve?

The Temporal API is already supported by Deno and the Boa JS engine.

https://github.com/tc39/proposal-temporal

What is the feature you are proposing to solve the problem?

Implement Temporal in Bun.

What alternatives have you considered?

Slow and heavy polyfills.

[kravetsone]

See temporalio/sdk-typescript#1334 ([Feature Request] Run temporal worker in bun).

Not the same temporal

[maurovanetti]

See temporalio/sdk-typescript#1334 ([Feature Request] Run temporal worker in bun).

Not the same temporal

This is embarrassing. Sorry. 😊

[TomasHubelbauer]

This would be great. It looks as though in Safari whose JavaScript engine, JavaScriptCore, Bun is using, has preview support with the useTemporal runtime flag: https://bugs.webkit.org/show_bug.cgi?id=223166 & https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Temporal#browser_compatibility

[notramo]

The patches contain just a stub, only 2 empty objects.

[kravetsone]

It would be great to have it
-1 dependency

[NuckChorris]

Looking upstream in JSC, despite the issue showing no movement in years, I see they've got an experimental implementation of Temporal: https://github.com/search?q=repo%3AWebKit%2FWebKit%20Temporal&type=code

I wonder if Bun could enable this experimentally like Deno did with the V8 Temporal implementation 👀

[Jarred-Sumner]

@NuckChorris try BUN_JSC_useTemporal=1: https://github.com/oven-sh/WebKit/blob/50ae7bfd503f896681ca5a9b848f9822c31c4912/Source/JavaScriptCore/runtime/OptionsList.h#L620

[notramo]

@Jarred-Sumner, isn't it only for the empty stub Temporal object? Does it enable a usable implementation?

[NuckChorris]

@notramo I would recommend running the Temporal Test262 Suite against it (if I find time this weekend I might run it and report back) but from the C++ it looks at least vaguely implemented-ish?

[jacksteamdev]

Implementation	Pass Rate	Execution Time	Notes
BUN_JSC_useTemporal	40%	0.8s
temporal-polyfill	88%	6s
@js-temporal/polyfill	84%	228s	Some >10s Duration tests

It's about 40% there... but the tests run blazing fast! The test runner README estimates ~30s to run, but Bun runs it in 0.8s on my M3.

[kravetsone]

Implementation	Pass Rate	Execution Time	Notes
BUN_JSC_useTemporal	40%	0.8s
temporal-polyfill	88%	6s
@js-temporal/polyfill	84%	228s	Some >10s Duration tests

It's about 40% there... but the tests run blazing fast! The test runner README estimates ~30s to run, but Bun runs it in 0.8s on my M3.

Wants it as much as possible sooner

Because it's really great api

[kravetsone]

nodejs/node#57128

NodeJS wants to unflag it

[jasnell]

Yes, we'd like to unflag it in node.js but still keep it marked as experimental, at least until v8's test coverage numbers are close to 100 and v8 chooses to expose it by default without the flag.

[jasnell]

The current thinking is that we likely will not actually merge the PR until after 24.0.0 is cut in April and keep it as experimental until v8's test coverage finally catches up.