Bug 99233 – Update the set-href-attribute-port test

• Component: WebCore Misc.
• Status: RESOLVED LATER

[Benjamin Poulain]

It's the test that turn out to be bad, not the parser.

[Benjamin Poulain]

Created attachment 168531 [details]
Patch

[Build Bot]

Comment on attachment 168531 [details]
Patch

Attachment 168531 [details] did not pass mac-ews (mac):
Output: http://queues.webkit.org/results/14292115

New failing tests:
fast/dom/HTMLAnchorElement/set-href-attribute-port.html

[Benjamin Poulain]

Created attachment 168534 [details]
Patch

[Benjamin Poulain]

Comment on attachment 168534 [details]
Patch

Never mind. According to http://dev.w3.org/html5/spec/single-page.html#dom-uda-protocol, KURL is actually correct. All the other implementations are incorrect.

[Ian 'Hixie' Hickson]

See also: http://url.spec.whatwg.org/
(ping Anne)

[Benjamin Poulain]

(In reply to comment #5)
> See also: http://url.spec.whatwg.org/
> (ping Anne)

Great!

Please also consider the case of HTMLAnchorElement which is quite a mess in practice.

[Anne van Kesteren]

Setting port should only work for URLs of which we know the scheme to support that: http://url.spec.whatwg.org/#hierarchical-scheme

What url.spec.whatwg.org calls URLUtils will eventually be how HTMLAnchorElement is defined. Just need to polish the specification a bit more and change a few terms around.

[Benjamin Poulain]

(In reply to comment #7)
> Setting port should only work for URLs of which we know the scheme to support that: http://url.spec.whatwg.org/#hierarchical-scheme
> 
> What url.spec.whatwg.org calls URLUtils will eventually be how HTMLAnchorElement is defined. Just need to polish the specification a bit more and change a few terms around.

This: http://url.spec.whatwg.org/#hierarchical-scheme is inconsistent with http://dev.w3.org/html5/spec/single-page.html#authority-based-url

It looks to me that http://url.spec.whatwg.org/#hierarchical-scheme just follows the principle of GoogleURL's IsStandardScheme() but with file: and filesystem: missing. (http://code.google.com/p/google-url/source/browse/trunk/src/url_util.cc)

Neither definition closely follows reality. What is the plan there, get rid of the URL part of the HTML5 spec?

[Anne van Kesteren]

Yes, HTML URL handling will go away in favor of the URL Standard. file URLs do not have a port either typically, but are usually hierarchical, indeed...

[Benjamin Poulain]

I'll fix that later as the new URL spec improve.