Bug 43246 – Web Inspector: incorrect absolute protocol-less URLs in tooltips of links in the ElementsTreeOutline

• Status: RESOLVED
• Resolution: FIXED
• Component: Web Inspector (Deprecated)

[Alexander Pavlov (apavlov)]

In a page loaded from http://example.com,
<a href="//example.org/foo">...</a> is given the following tooltip in the ElementsTreeOutline:
"http://example.com//example.org/foo", which should be "http://example.org/foo" instead.

[Alexander Pavlov (apavlov)]

Created attachment 63333 [details]
[PATCH] Suggested solution

[Joseph Pecoraro]

Hmm, I didn't even know about this style of link. Is it specified somewhere?
I took a look at the HTML5 spec and nothing popped out at me. I'd appreciate
a link if have one.

Also, I notice a number of calls to completeURL (in frames) just use the
documentURL. Does that respect a <base> element if there is one? Meaning,
does the documentURL property update if there is a <base href="..."> or
xml:base attribute, or must those be determined dynamically?

[Alexander Pavlov (apavlov)]

(In reply to comment #2)
> Hmm, I didn't even know about this style of link. Is it specified somewhere?
> I took a look at the HTML5 spec and nothing popped out at me. I'd appreciate
> a link if have one.

Please see http://www.faqs.org/rfcs/rfc1808.html (2.4.3) for the reference.

> Also, I notice a number of calls to completeURL (in frames) just use the
> documentURL. Does that respect a <base> element if there is one? Meaning,
> does the documentURL property update if there is a <base href="..."> or
> xml:base attribute, or must those be determined dynamically?

I'm not entitled to answer this question :) The only thing I've done in this direction was the introduction of the documentURL property which is determined in InspectorDOMAgent as document->url(). I do not know whether the parser takes <base href...> into account when computing document->url() but anyway, I believe it is the right point where this should be done.

[Joseph Pecoraro]

Comment on attachment 63333 [details]
[PATCH] Suggested solution

> (use the baseURL protocol).

NIT: might be better if it was "(use the baseURL's protocol).".


> Please see http://www.faqs.org/rfcs/rfc1808.html (2.4.3) for the reference.

Thanks, this is great =). I see this format under Relative URLs.


> > Also, I notice a number of calls to completeURL (in frames) just use the
> > documentURL. Does that respect a <base> element if there is one? Meaning,
> > does the documentURL property update if there is a <base href="..."> or
> > xml:base attribute, or must those be determined dynamically?
> 
> I'm not entitled to answer this question :)

I think "entitled" is the wrong word. But I understand what you mean =).


> The only thing I've done in this direction was the introduction of the documentURL property
> which is determined in InspectorDOMAgent as document->url(). I do not know whether the
> parser takes <base href...> into account when computing document->url() but anyway, I
> believe it is the right point where this should be done.

I just whipped up a quick test, and the Inspector's tooltips do not work correctly with
<base>. I just filed the following:

  Web Inspector: Relative URL Link Tooltips do not respect <base>
  https://bugs.webkit.org/show_bug.cgi?id=43434

Let me know if its something you are interested in working on. The thing is so few people
use <base> that it might not be a good use of time. Still, it is a bug.

r=me for this fix. Thanks!

[Alexander Pavlov (apavlov)]

Committing to http://svn.webkit.org/repository/webkit/trunk ...
        M       WebCore/ChangeLog
        M       WebCore/inspector/front-end/inspector.js
Committed r64565