- From: Henri Sivonen <hsivonen@iki.fi>
- Date: Mon, 12 Apr 2010 15:55:22 +0300
- To: Ian Hickson <ian@hixie.ch>
- Cc: HTMLWG WG <public-html@w3.org>
On Jul 28, 2009, at 09:24, Ian Hickson wrote: > On Mon, 13 Jul 2009, Henri Sivonen wrote: >> >> Could you please check if the specced behavior for >> http://hixie.ch/tests/adhoc/dom/level0/write/005.html is the wanted >> behavior and then write down what the wanted result is? (I'm not sure I >> got it right myself.) > > I've updated the test to match what the spec says (and to more > dramatically demonstrate why I think what it says is substantially > preferable than what WebKit and Gecko do today). Currently, in Gecko nightlies with the HTML5 parser enabled, I get: "FAIL (class attribute present - appendChild()ed script took too long to run) - PASS" and the document never finishes loading. The script says "else if (!document.getElementsByTagName("P")[0].hasAttribute('class')) s = 'FAIL (class attribute present - appendChild()ed script took too long to run)';" The negation in the |if| condition doesn't match the message. Furthermore, I don't see any part of the test setting or removing the class attribute. Is the test now in the state that it is supposed to be in? >> Also, have I understood correctly that insertion point is never >> undefined when there's an active parser that is a script-created parser >> and the tokenizer isn't tokenizing the synchronous part of a >> document.write()? > > Not sure what you mean by the second part. The insertion point is never > undefined when there's an active script-creaed parser full stop. It's not necessarily good to let the SVG load event or XBL constructors to call document.write() even in the script-created case. -- Henri Sivonen hsivonen@iki.fi http://hsivonen.iki.fi/
Received on Monday, 12 April 2010 12:55:59 UTC