Filed in response to bug 18333 comment 60. Code from HTML sink contexts was forked to the XML content sink as part of the fix for bug 18333. Ideally, this code should be unforked and moved to a superclass. However, that probably cannot happen with sink contexts in place on the HTML side. (It seems to me that rewriting the HTML parser and content sink to implement the HTML5 parsing algorithm would give a good opportunity to address this issue.)

I guess this now means replacing nsXMLContentSink with something that emits nsHtml5TreeOperations that feed into nsHtml5TreeOpExecutor.

Andrew, this is actually a good tech debt item to deal with for DOM; there's a bunch of code that can go away if we do this....

I am also exploring alternatives to move this forward, setting NI on me as a reminder.

Alphan will start to study this in Q4.

After discussing with Henri, this bug should be XML rewrite. The bug will make the XML Web content load path have the same structure as the HTML loads path. nsXMLContentSink would be replaced by mozilla::parser::xml::TreeOpGenerator(new added) and nsHtml5TreeOperations/nsHtml5TreeOpExecutor(eventually to be named mozilla::parser::TreeOperations/TreeOpExecutor). The detailed work is described in the following link. https://wiki.mozilla.org/Platform/XML_Rewrite At first, I will try to resolve bug 717561, bug 737778 and bug 737845. One more thing, the incrementalism is not worked currently on the two test pages. http://www.damowmow.com/mozilla/bugs/18333/xml a test served as text/xml http://www.damowmow.com/mozilla/bugs/18333/html same, served as text/html

Rename this bug to match the current goal.