Expose URLPattern everywhere

[ptomato]

I've drafted a set of criteria for which interfaces to expose universally on all globals, at
w3ctag/design-principles#510. URLPattern looks like it fits those criteria and is useful to have in any environment.

• At least two implementers are interested (and none opposed):
  • …
  • …
• Tests are written and can be reviewed and commented upon at:
  • Enable URLPattern tests in ShadowRealm web-platform-tests/wpt#49323
• Implementation bugs are filed:
  • Chromium: …
  • Gecko: …
  • WebKit: …
  • Deno: …
  • kenchris/urlpattern-polyfill: N/A
• MDN issue is filed: …
• The top of this comment includes a clear commit message to use.

(See WHATWG Working Mode: Changes for more details.)

---

Preview | Diff

[jeremyroman]

Conceptually makes sense to me, but would like to see w3ctag/design-principles#510 settled.

Browsers (well, at the moment, Chromium) shipping this probably does imply a web-facing change albeit a minor one.

Does this confine the future evolution of URLPattern? For example, it says that anything that uses an event loop or I/O should not have [Exposed=*]. Would we be deciding, now, that no future members like that would be added (or, at least, that they would throw)? I don't anticipate anything like that, but want to be sure I understand all of the consequences.

[ptomato]

Thanks! I will proceed with writing tests and filing implementation bugs when the design-principles issue is settled, then.

I don't believe this would prevent adding members in future that wouldn't be exposed everywhere. For example, AbortSignal has a timeout member that is only exposed in Window and Worker, despite AbortSignal having [Exposed=*]. And if there's a good reason and all else fails, well, the design principle is a guide, not a requirement.

[ptomato]

@jeremyroman The design-principles issue is settled (https://www.w3.org/TR/design-principles/#expose-everywhere) so I think this can be moved forward. What should the next steps be? From my side, a review of the tests in web-platform-tests/wpt#49323 would be great.

[jeremyroman]

Makes sense to me for this to be Exposed=* for the reasons you gave. I'm happy to merge this if you can step through the rest of the checklist. It looks like this is already there for Chromium (at least, up to a TODO that applies to other things that have Exposed=* in the spec). Left an initial review on the WPT PR you linked.

[sisidovski]

Still catching up with the context but generally the change sounds reasonable.

@ptomato Do you have other ongoing (or merged) PRs to expose some APIs everywhere? e.g. the design principle says TextEncoder is a good example.

[ptomato]

@sisidovski Yes, here are the others:

• webcrypto: Allow using Crypto.getRandomValues() in Shadow Realms w3c/webcrypto#361
• console: Expose everywhere console#209
• events, AbortController, AbortSignal: Expose events/aborts everywhere dom#1024
• TextDecoder, TextEncoder: Expose everywhere encoding#274
• ErrorEvent, PromiseRejectionEvent: Expose Error/PromiseRejectionEvent everywhere html#7244
• streams: Expose everywhere streams#1176
• URL: Expose URL/URLSearchParams everywhere url#657
• compression: Expose everywhere compression#40
• wasm: [js-api] Expose everywhere WebAssembly/spec#1550
• definition of UniversalGlobalScope which is the global scope object equivalent of Exposed=*; there's much more in that PR, so I linked to the relevant section

[sisidovski]

Thanks! LGTM.

I think you can fill the checkbox in the description?