- From: Boris Zbarsky <bzbarsky@MIT.EDU>
- Date: Thu, 14 Apr 2011 09:46:24 -0400
- To: HTML WG <public-html@w3.org>
Consider the following testcase: <!DOCTYPE html> <form name="x"> <img name="y"> </form> <script> alert(document.forms["x"].y); alert(document.forms["x"].elements.y); </script> The results are as follows: IE9: alerts undefined twice (in both standards and quirks mode as far as I can tell). Opera 11: alerts [object HTMLImageElement], then undefined Firefox: alerts [object HTMLImageElement], then undefined Chrome: alerts [object HTMLImageElement] twice Safari: alerts [object HTMLImageElement] twice Chrome and Safari are consistent with what the spec currently says for the namegetter on HTMLFormElement, but not with the behavior specified for form.elements. Opera and Firefox do what the spec currently says for form.elements, but not for HTMLFormElement. IE currently does what the spec currently says. I'm pretty sure I've seen code in the past that switches on "IE vs Netscape" and uses form.x for the latter while using getElementById for the former, but that was a good long while ago. So I think we're willing to risk the compat hit in Gecko of changing to the currently specced behavior (which will incidentaly be simpler than what we do now). Are Opera and WebKit willing to do that too? Or does the spec need to change? -Boris
Received on Thursday, 14 April 2011 13:46:52 UTC