RE: [SVGMobile12] namespace events are not qualified?

HI Jon,
 
My suggested resolution though, is to change DOM 3 Events. I would say that
it is completely unworkable as it stands, not just for SVG but for any
application that makes use of both DOM 2 and DOM 3 Events.
 
The main problem from a backwards compatibility problem is that I cannot
generate an event "xyz:abc" safe in the knowledge that only listeners who
*really* want that event will be triggered. I cannot be sure that I won't
also trigger listeners for "abc", who have never even heard of DOM 3 Events,
or the set of events that sit in the namespace "xyz".
 
Regards,
 
Mark
 

Mark Birbeck
CEO
x-port.net Ltd.

e: Mark.Birbeck@x-port.net
t: +44 (0) 20 7689 9232
w: http://www.formsPlayer.com/
b: http://internet-apps.blogspot.com/

Download our XForms processor from
http://www.formsPlayer.com/ 

 


  _____  

From: www-svg-request@w3.org [mailto:www-svg-request@w3.org] On Behalf Of
Jon Ferraiolo
Sent: 25 May 2005 01:07
To: Mark Birbeck
Cc: 'KONO Masahiko'; www-svg@w3.org
Subject: RE: [SVGMobile12] namespace events are not qualified?


Mark,
[As usual!] You are absolutely correct in your analysis. Thanks for
carefully describing this whole process and how events are now promiscuous
(i.e., DOM2Events "click" really means DOM3Events {"*", "click"}). I suggest
that the SVG-t 1.2 spec (and also the SVG-Full 1.2 spec) highlight this
somehow to make sure implementers and developers don't get confused (as I
was).

Also, I think there needs to be some spec changes in section 13.2
[http://www.w3.org/TR/SVGMobile12/interact.html#SVGEvents]. For example, the
SVG-t 1.2 spec says:

-------------------------------------- 

*	SVG 1.1 names are all assumed to be in the
"http://www.w3.org/2001/xml-events" namespace. This allows SVG 1.1 content
(which did not have a notion of namespaced events) to be upwardly compatible
with SVG 1.2 (which adds a notion of namespaced events). Therefore, the SVG
1.1 "SVGZoom" event becomes the {"http://www.w3.org/2001/xml-events",
"SVGZoom"} event in SVG 1.2. 

--------------------------------------

Based on the promiscuity statements about backwards compatibility with DOM2
Events, I believe the above bullet is wrong. The old "SVGZoom" is really
{"*", "SVGZoom"}.

Jon

At 05:26 PM 5/19/2005, Mark Birbeck wrote:


Hi Jon,

> Very important issue here.

Definitely.


>  Thanks for initiating discussion.

Thanks for replying so quickly ;)


I said:
> >             Let's say we register for:
> >             
> >                * click
> >                * ev:click
> >                * myns:click
> >                * yourns:click
> >             
> >             
> >             1. If we dispatch the old DOM 2 Event "click", *all* of
these listeners
> >                will be triggered.

You replied:
>       Assuming the ev: is assigned to the XML Events namespace and myns:
and yourns:
> are assigned to different namespaces, then, no, if the mouse is clicked,
then only
> the event listeners for "click" and "ev:click" would receive the event.
At least
> that is my thinking based on studying the DOM3 Events spec which appears
to retroactively
> claim that HTML's click event when namespaced becomes ev:click.

No, I'm pretty certain that this is not what happens. As you rightly say,
the implication of the DOM 3 Events spec is that there is no longer a
"click" event, but there is now a {"http://www.w3.org/2001/xml-events",
"click"} event. So far so good.

However, it also says that if the *listener* is registered for an event with
no namespace (e.g., "click") OR the event generated has no namespace (e.g.,
{null, "click"}), then the types will match, *regardless* of the namespace
[1]:

   For compatibility reasons, the dispatching of an event will ignore
namespace URIs
   if either the event or the event listener has a null namespace URI.

The Note goes on to describe what this means:

   If a DOM Level 2 event (i.e. with a null namespace URI) is dispatched in
the DOM
   tree, all event listeners that match the type will be triggered as
described in
   the DOM event flow.

So, say I give you an XBL document to use with your documents, and my
document is written to use DOM 2 Events. If it dispatches good old-fashioned
"click", *all* of the following DOM 3 Event listeners will be triggered in
your document:

   * ev:click
   * myns:click
   * yourns:click

as well as any DOM 2 Event listeners waiting for "click". Note that this
means that it's not enough for you to avoid using type names from the list
of 'core' events -- the problem is caused by any DOM 2 Event.

So, just so that everything is clear, if you don't mind me
cutting-and-pasting your words, I would say that when you say this:

  ...if the mouse is clicked, then only the event listeners for "click" and
  "ev:click" would receive the event...

although it's what we would like, the reality is you can't say "only":

  ...if the mouse is clicked, then the event listeners for "click" and
  "ev:click" would receive the event, as would any other listener with
  "*:click"...


Next the Note says this:

   If a DOM Level 3 event (i.e. with a namespace URI) is dispatched in the
DOM tree,
   all event listeners with the same type and the same or null namespace URI
will be
   triggered as described in the DOM event flow.

The consequence of this is that if my XBL object is 'listening' for "click"
to come from your document, any handlers that I have registered will be
invoked if my code receives *any* of the following events from your
document:

   * ev:click
   * myns:click
   * yourns:click

as well as the DOM 2 Event "click".

In other words we have {"*", "click"}, if you'll forgive my notation.

        
>       What I am assuming is that an implementation would create a DOM3
Event object
> for all events, whether the event listener uses DOM2 or DOM3 methods. For
"click"
> events, the namespaceURI value on the Event object would be set to
> "http://www.w3.org/2001/xml-events".

In DOM 3 Events-compliant code it would, since you would use initEventNS().
But sticking with my example, my DOM 2 Event-compliant code will call
initEvent() on this Event object, thus setting the namespace URI to null. If
I then call dispatchEvent() on some target node, it would of course
correctly trigger any listener for "click". But as I've said, it would also
trigger any other listeners with a type of "click", even if they have
non-null namespaces.

And further, I'm saying that the "click" listener that you set up to listen
for this "click" event would also be triggered by an event that was
initialised with initEventNS() and dispatched as {"myns", "click"}.

To put all of this a different way, it is only fully qualified event names
(i.e., QNames) that are 100% unambiguous. Any use of unqualified names is
ambiguous.

[snip]


>       Actually, I think it all works nicely if "click" is truly
equivalenced to
> "ev:click", meaning that the assumed default namespace for legacy HTML and
SVG
> events is "ev:".

But as I've tried to show, this is actually not possible to achieve -- there
is no way I can see in the DOM Events model to make this kind of
'equivalence' (unless the core events are in no namespace in both DOM 2
Events and DOM 3 Events).


> I believe that attempting to wildcard "click" to mean everything:click is
what
> would cause problems.

Exactly!


I have to say, I cannot see how it is possible to say that "click" ==
"ev:click" if you are allowing DOM 2 Events and DOM 3 Events to co-exist.
The whole "ignore namespaces" thing is an attempt to solve this problem, but
as far as I can see it just doesn't work.

(I think there are other problems with DOM 3 Events, but I'll leave those
for a separate thread.)

Regards,

Mark

[1]
http://www.w3.org/TR/DOM-Level-3-Events/events.html#Events-dom2-compatibilit
y


Mark Birbeck
CEO
x-port.net Ltd.

e: Mark.Birbeck@x-port.net
t: +44 (0) 20 7689 9232
w: http://www.formsPlayer.com/
b: http://internet-apps.blogspot.com/

Download our XForms processor from
http://www.formsPlayer.com/