Bug 115755

Summary: Devirtualize Document class type checking
Product: WebKit Reporter: Ryosuke Niwa <rniwa>
Component: DOMAssignee: Ryosuke Niwa <rniwa>
Status: RESOLVED FIXED    
Severity: Normal CC: benjamin, commit-queue, d-r, eric.carlson, esprehn+autocc, fmalita, glenn, japhet, jer.noble, kling, koivisto, pdr, schenney
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
Attachments:
Description Flags
Cleanup benjamin: review+

Description Ryosuke Niwa 2013-05-07 13:50:48 PDT
Merge
https://chromium.googlesource.com/chromium/blink/+/dae5adc768d5ce6bff301df6515745da8ea24950
Devirtualize Document class type checking

Document has a bunch of virtual bool is*Document() methods on it, but it also has
two bools for XHTML and HTML documents which is silly. We can merge them all
together into an enum of values and devirtualize the type checking methods.

This patch introduces the DocumentClass enum that is used to signal what subclass
of Document, or HTMLDocument, is being implemented.
Comment 1 Ryosuke Niwa 2013-05-07 16:26:00 PDT
Created attachment 200991 [details]
Cleanup
Comment 2 Benjamin Poulain 2013-05-07 16:37:54 PDT
Comment on attachment 200991 [details]
Cleanup

View in context: https://bugs.webkit.org/attachment.cgi?id=200991&action=review

> Source/WebCore/dom/Document.h:1486
> +    unsigned m_documentClasses;

I would just use a uchar8_t here. That way it fits on a word with the 3 boolean bellow.

> Source/WebCore/html/HTMLDocument.h:78
> +    HTMLDocument(Frame*, const KURL&, unsigned documentClasses = HTMLDocumentClass);

I think it should have a default of 0.

Otherwise, an caller may think passing something else override the HTMLDocumentClass. Which is not true because the constructor does the right thing.
Comment 3 Ryosuke Niwa 2013-05-07 18:59:29 PDT
Committed r149705: <http://trac.webkit.org/changeset/149705>