[Webkit-unassigned] [Bug 181460] REGRESSION(r225650): The scores of MotionMark tests Multiply and Leaves dropped by 8%

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Thu Jan 11 15:10:19 PST 2018


https://bugs.webkit.org/show_bug.cgi?id=181460

--- Comment #17 from Chris Dumez <cdumez at apple.com> ---
Comment on attachment 330906
  --> https://bugs.webkit.org/attachment.cgi?id=330906
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=330906&action=review

>>> Source/WebCore/css/parser/CSSParser.cpp:81
>>> +    , hasDocumentSecurityOrigin(document.baseURL() == baseURL || document.securityOrigin().canRequest(baseURL))
>> 
>> Can this tweak be made in canRequest to make it fast instead?
>> 
>> But also, I'm actually not sure if identical base URL is good enough for not having a security boundary, between origins in this case.
> 
> As for the first question, maybe there is some other way to make a fast path, but we can’t make this exact optimization because canRequest does not have access to the document’s baseURL.
> 
> For the second question, I suggest we work together to turn your uncertainty into a test case.

I would worry a lot less if this would be using document.url() rather than document.baseURL(). AFAIK, the document baseURL can be overridden by a <base> element and the document's securityOrigin comes from its URL, not its baseURL.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20180111/89644b20/attachment-0001.html>


More information about the webkit-unassigned mailing list