Use the url spec to parse ice server urls. #2694
Conversation
webrtc.html
Outdated
| <p>If <var>scheme</var> is <code>stun</code> or | ||
| <code>stuns</code>, and <var>parsedURL</var> violates | ||
| the syntax defined in [[!RFC7064]] section 3.1, | ||
| then [= exception/throw =] a {{SyntaxError}}.</p> |
There was a problem hiding this comment.
This doesn't work as parsedURL is a data structure. How is this implemented?
There was a problem hiding this comment.
This tries to preserve the existing normative language. I shold s/parsedURL/url/ again perhaps. As I mention in the OP, looking for ideas on how to specify the requirements using the URL spec. We want to prevent hierarchical urls, e.g. stun:foo.bar.com:1234
There was a problem hiding this comment.
In Firefox, this is implemented here.
There was a problem hiding this comment.
That URL isn't hierarchical though, is it? Hierarchical URLs contain slashes, as far as I know.
It sounds like you want https://url.spec.whatwg.org/#url-opaque-path to be true, although you wouldn't have to check for credentials in that case.
And then as per the Firefox code you'd have to parse parsedURL's path (which will be a string in this case) in a custom manner to obtain the host and port (split on :, then handle each?). Potentially this logic could be shared with the URL Standard in some way.
And then you'd handle the parsedURL's query as per Firefox logic for the query.
|
@dontcallmedom The failing infra checks here seem unrelated, and appear from DOMTimeStamp being no more. Should we perhaps merge #2686 until #2674 has been resolved? Otherwise we appear blocked from submitting any PRs on this repo, unless there's some workaround to suppress the errors? |
|
Closing in favor of #2853. |
Fixes #2660 (draft).
@annevk I'm open to ideas on how to iterate further while at the same time capturing the normative requirements in rfc7064 and rfc7065, such as "Developers MUST NOT use a generic hierarchical URI parser to parse a "stun" or "stuns" URI."
Preview | Diff