- From: Michael A. Puls II <shadow2531@gmail.com>
- Date: Sat, 27 Mar 2010 19:08:17 -0300
- To: "Adam Barth" <w3c@adambarth.com>
- Cc: public-html@w3.org
On Sat, 27 Mar 2010 19:09:25 -0300, Adam Barth <w3c@adambarth.com> wrote:
> On Sat, Mar 27, 2010 at 12:47 PM, Michael A. Puls II
> <shadow2531@gmail.com> wrote:
>> On Fri, 26 Mar 2010 21:47:35 -0300, Adam Barth <w3c@adambarth.com>
>> wrote:
>>> 2) Specify the Firefox behavior in HTML5 and implement it in WebKit.
>>
>> In the async case:
>>
>> What do you want to happen with load("file not found")? Do you want
>> load()
>> to throw an exception like Firefox, or do you want it to not throw an
>> exception, still fire the 'load' event and just have
>> event.target.documentElement be null like Opera?
>>
>> What do you want to happen when load("file is found") is an xml file
>> with a
>> parse error? In the 'load' callback, do you want just
>> event.target.documentElement to be null, or do you want documentElement
>> to
>> be a 'parsererror' element in the
>> "http://www.mozilla.org/newlayout/xml/parsererror.xml" namespace like
>> Firefox does (basically a yellow screen of death document). How do other
>> vendors feel about using that ns string?
>>
>> If load() does throw an exception, could it perhaps be standardized on
>> the
>> name of the exception? Firefox uses something like NS_ERROR_DOM_BAD_URI
>> or
>> something.
>>
>> It seems you can call load() multiple times on the same object and it'll
>> work just fine too. Not sure if that's needed, but mentioning it.
>>
>> Also, with the sync version at least, does load() return true|false like
>> Firefox, or is it a void function like in Opera and returns undefined?
>>
>> How do you trigger the sync version? Is it doc.async = false? Or, is it
>> the
>> lack of an onload property or registered 'load' listener that triggers
>> the
>> sync version?
>>
>> Please clarify 'Firefox behavior' in regards to the above. Do you want
>> to
>> copy Firefox exactly 100%?
>
> All good question, which we can figure out once we decide on an
> approach. The synchronous behavior is triggered by setting
> document.async to false.
Just to add:
Should 'load' still fire if doc.async = false? It does in Firefox, but not
Opera.
Also, in Opera, if doc.async = true, load() returns null. If doc.async =
false, load() returns an instanceof Element (not XMLElement). Firefox
returns true.
--
Michael
Received on Saturday, 27 March 2010 23:08:51 UTC