Bug 213167 – [WPT] dom/nodes/Document-createCDATASection-xhtml.xhtml fails due to missing exception in Document.createCDATASection()

• Component: DOM
• Status: RESOLVED FIXED

[Sam Weinig]

WPT: dom/nodes/Document-createCDATASection-xhtml.xhtml fails due to missing exception in Document.createCDATASection()

Test: https://wpt.fyi/results/dom/nodes/Document-createCDATASection-xhtml.xhtml?run_id=586650004&run_id=579020002&run_id=594530003&run_id=583030005
Spec: https://dom.spec.whatwg.org/#dom-document-createcdatasection

Issue: We don't implement step 2 of "createCDATASection(data)".

[Sam Weinig]

Created attachment 401841 [details]
Patch

[Yusuke Suzuki]

Comment on attachment 401841 [details]
Patch

r=me

[Sam Weinig]

Created attachment 401843 [details]
Patch

[EWS]

Committed r263003: <https://trac.webkit.org/changeset/263003>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 401843 [details].

[Radar WebKit Bug Importer]

<rdar://problem/64330801>

[Radar WebKit Bug Importer]

<rdar://problem/64330800>

[Alexey Proskuryakov]

Comment on attachment 401843 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=401843&action=review

> Source/WebCore/dom/Document.cpp:1031
> +    if (data.contains("]]>"))

Does cdatasection::create not need the check?

Performance implications seem non-trivial here.

[Sam Weinig]

(In reply to Alexey Proskuryakov from comment #7)
> Comment on attachment 401843 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=401843&action=review
> 
> > Source/WebCore/dom/Document.cpp:1031
> > +    if (data.contains("]]>"))
> 
> Does cdatasection::create not need the check?

I don't believe it does. The only other caller of CDATASection::create (other than right after this check) is XMLDocumentParser::cdataBlock, and I don't think the parser can get into this issue.

> 
> Performance implications seem non-trivial here.

Really? I don't think of something like document.createCDATASection() getting called all that much from the DOM. We now match the other major browsers and the spec, so it does turn out to a perf issue, we should also raise it with the standards folks.

[Chris Dumez]

(In reply to Sam Weinig from comment #8)
> (In reply to Alexey Proskuryakov from comment #7)
> > Comment on attachment 401843 [details]
> > Patch
> > 
> > View in context:
> > https://bugs.webkit.org/attachment.cgi?id=401843&action=review
> > 
> > > Source/WebCore/dom/Document.cpp:1031
> > > +    if (data.contains("]]>"))
> > 
> > Does cdatasection::create not need the check?
> 
> I don't believe it does. The only other caller of CDATASection::create
> (other than right after this check) is XMLDocumentParser::cdataBlock, and I
> don't think the parser can get into this issue.
> 
> > 
> > Performance implications seem non-trivial here.
> 
> Really? I don't think of something like document.createCDATASection()
> getting called all that much from the DOM. We now match the other major
> browsers and the spec, so it does turn out to a perf issue, we should also
> raise it with the standards folks.

I agree with Sam here.