Re: Updated/split test

On Mon, 27 Oct 2008 16:30:50 +0100, Niklas Hagelroth <niklas.hagelroth@ikivo.com> wrote:

>
>
> Erik Dahlström wrote:
>> On Mon, 27 Oct 2008 14:31:24 +0100, Niklas Hagelroth
>> <niklas.hagelroth@ikivo.com> wrote:
>>
>> >
>> > Erik Dahlström wrote:
>> >> <niklas.hagelroth@ikivo.com> wrote:
>> >>
>> >> > Hi,
>> >> >
>> >> > Yes, Ikivo pass struct-use-205 revision 1.6.
>> >> >
>> >> > I'm still not convinced that the 4th subtest that is moved to struct-
>> >> use-210 is corrected. Can you please point me to the correct section in
>> >> the specification that proves this test correct?
>> >>
>> >> I maintain that the struct-use-210 test is valid (even if it may be a
>> bit
>> >> tricky), and it was reviewed by a second WG member who also found it
>> valid.
>> >>
>> >> > Also, since the test is so complicated and hard to find evidence for
>> in
>> >> the specification I would like to see links to the tested specification
>> >> sections in the tests description.
>> >>
>> >> The normative definition for SVGElementInstance.correspondingUseElement
>> is
>> >> in SVG 1.1, which is linked to from the uDOM[1]. Also take note of the
>> SVG
>> >> 1.1 errata about liveness of DOM properties, which may play a part in
>> this.
>> >
>> > Since the chain of EventTargets is determined before the initial
>> dispatch of the event, changing the tree will not affect the event
>> processing. So referring to the liveness of the DOM seems irrelevant in
>> this case.
>>
>> The EventTarget chain is determined before the Event is dispatched, as per
>> DOM 2 Events. That has nothing to do with
>> SVGElementInstance.correspondingUseElement however
>> (correspondingUseElement is what the test is meant to test). The test does
>> not in any way affect the Event object directly, or the way the event
>> bubbles. The properties on a particular SVGElementInstance (which may be a
>> target/currentTarget in the Event object) are subject to 'live' updates to
>> match the current DOM tree.
>
> If it is the case (that it was subject to live updates) it seems strange that the object that event.target points to exists at all, since the SVGElementInstance object shouldn't exist if the element is not <use>d anymore.

We agreed a few lines up that the EventTarget chain was determined before the event was dispatched (DOM 2 Events), so I don't see how you can argue now that it needs to be broken.

Cheers
/Erik

-- 
Erik Dahlstrom, Core Technology Developer, Opera Software
Co-Chair, W3C SVG Working Group
Personal blog: http://my.opera.com/macdev_ed

Received on Monday, 27 October 2008 17:10:27 UTC