Fix unexpected stripping of default port in port canonicalization

[sisidovski]

This prevents the https default port (443) from being replaced with the empty string in canonicalize a port. The algorithm creates an https dummy URL (https://dummy.invalid/), any partial substring that happens to match 443 (e.g., the "443" in new URLPattern({ port: "*443" })) gets interpreted as the default port for https and is wrongly stripped to the empty string.

This modifies the canonicalize a port algorithm so that when it is invoked without an explicit protocolValue, it sets the dummy URL's scheme to the empty string rather than leaving it as "https".

• At least two implementers are interested (and none opposed):
  • Chrome
  • …
• Tests are written and can be reviewed and commented upon at:
  • web-platform-tests/wpt@23aac92
• Implementation bugs are filed:
  • Chromium: https://crbug.com/520829534
  • Gecko: …
  • WebKit: …
  • Deno: …
  • kenchris/urlpattern-polyfill: …
• MDN issue is filed: N/A
• The top of this comment includes a clear commit message to use.

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

Fixes #285.

---

Preview | Diff

[annevk]

This is a normative change so we should use the PR template, no?

[shannonbooth]

I went digging, apparently the test for this has already been merged: web-platform-tests/wpt@23aac92

edit: and FWIW this MR gets that test working for my implementation

[sisidovski]

Updated the description to use PR template.

Similar to #272, this is more like a spec bug after 46c30fd. We've confirmed that both Safari and Firefox have already implemented the behavior consistent with this change.