Missing use cases for stopImmediatePropagation event ordering (was: Re: ACTION-70: Define the scope chain of onFoo events reference issue-1)

"Bjoern Hoehrmann" <derhoermi@gmx.net>
> So if you think we should
> not define ordering, I'd rather hear why the downsides of defining
> the order outweight the cited benefits

cited benefits?  The one that relates to stopImmediatePropogation is the 
only one given, I find it hard to say what the downsides are to things which 
haven't been given, like most stuff, there are no use cases. If we talking 
ideal specifications that contain everything everyone wants minutely 
specified in detail, then I'd like to see actual use cases in the 
specification, rather than the current nowhere at all.

> and how similar effects can be achieved e.g. in XBL2 where
> stopImmediatePropagation seems to be
> pretty much required. In a new thread, ideally.

I don't see the need to pre-empt what the Web Application Formats group 
might produce in the area, much better to wait until they produce 
requirements and then meet them (or why not just let them meet them 
themselves)  as I've said previously I can't see the value in an ordering 
unless you also have the ability to insert first in the order - without 
either introspection or a specific method to do that, it's impossible, so I 
believe after the remove introspection decision from the f2f , it's 
impossible, this doesn't seem a good thing, as you can't actually know that 
the event you want to stopImmediatePropagation on is actually the first 
event listener attached, so you're still down to relying on the fact you've 
authored in a particular way.

Without knowing the use cases for the various features I am not able to 
review if the features meet them, this has been a long problem, of course 
normally it's okay as I can just ignore them and assume their okay.  If you 
want me to review DOM 3 events in its entirety, can I see some use cases? 
Maybe they're all in member space.

stopImmediatePropagation is trivial to implement outside of XBL situations- 
as is event ordering, you simply don't attach multiple events but have just 
one listener - if the events are really so intertwined that they must be 
handled together then it's not obvious why an author would attach 2 
listeners.

This is of course not the case in a XML componentised language such as XBL2, 
but then the current DOM 3 events spec -groups already meet this use with 
the ordering defined _within a particular group_.  I'm happy with that, it 
seems to meet this requirement  of course like I say, I only know of it from 
a technical requirement - use cases are wholly missing.

Jim. 

Received on Thursday, 16 March 2006 21:00:19 UTC