Currently, when rendering an XHTML page, WebKit does not recognize character entity references from the HTML namespace unless the MIME Type is application/xhtml+xml (or obviously text/html). If the page is delivered as application/xml, the named character entity references do not work. A better behavior (and the one supported in FireFox), would be to resolve and render the character entity referrences based on the namespace(s) of the document fragment. In this way simply declararing the namespace on a document fragment would enable charaqcter entity references from that namespace for that fragment: at least for those namespaces WebKit had knowledge of.
Created attachment 13749 [details] listing of named, numeric (hex), and numeric (dec) character references
Created attachment 13751 [details] listing of named, numeric (hex), and numeric (dec) character reference (for comparison)
I'm going to work on some more test cases. It appears these attachments load fine. Whieh may mean this is releated to an XSLT. It appears the bug is invalid. Could someone make that change. I'll change the it back or reopen if I learn anything new.
Marking as RESOLVED/INVALID per Comment #3.
Please also note that entity substitution happens earlier than the namespace becomes known. After all, entities are supposed to work in XML attributes themselves, too.
(In reply to comment #5) > Please also note that entity substitution happens earlier than the namespace > becomes known. After all, entities are supposed to work in XML attributes > themselves, too. > However, that would only be an issue if the xmlns defined additional charaacter entity references (which I believe it does not). Even if it does, it just means that the attributes in that xmlns namesapce would have to be processed first before the branches of the tree within those namespace declarations could be processed.