W3C home > Mailing lists > Public > www-dom@w3.org > January to March 2002

Re: SVG WG is asking for feedback - event propagation

From: Ray Whitmer <rayw@netscape.com>
Date: Fri, 22 Mar 2002 10:13:19 -0800
Message-ID: <3C9B743F.8020208@netscape.com>
To: Joseph Kesselman <keshlam@us.ibm.com>
CC: Thierry Kormann <tkormann@ilog.fr>, W3c-Svg-Wg <w3c-svg-wg@w3.org>, Www-Dom <www-dom@w3.org>, Arnaud Le Hors <lehors@us.ibm.com>, Philippe Le Hégaret <plh@w3.org>
It seems that such a bridge would have to be created by the outer layer 
and declare a handler on the inner layer.  What seems to be missing to 
me is the way for the outer layer to discover the document node of the 
embedded layer so it could register such a thing.  I do not think that 
seperate trees should be automatically connected.  Such a mechanism 
should be carefully considered.  In some cases it would violate security 
to permit them to even be manually bridged, because since the trees were 
loaded from seperate URIs, it could be from seperate domains and 
bridging them could lead to one domain being able to appropriate the 
trust of another.

Ray Whitmer

Joseph Kesselman wrote:

>The DOM hasn't specified event propigation between document trees. It looks
>like it would be easy to implement bubbling, at least,  by simply
>installing an event handler on the root of the referenced tree that
>"bridges" (re-posts) the event to the referencing tree's <image/> node.
>Capture may be more difficult to emulate, mostly because (as far as I can
>remember) the event target doesn't tell us whether a capture occurred and
>decided to block further processing, so the "bridge" handler doesn't know
>what to do about captures lower in the referenced tree. If this is an
>issue, the question becomes one of whether it should be handled by having
>SVG extend the events APIs or by doing that at the DOM level.
>But for bubbling, having SVG create a bridge handler when it builds the
><image/> reference seems to work perfectly well and produce the expected
>results (inner document sees the event before outer, and can cancel further
>processing, just as if it has been inserted at this point in the document
>Joe Kesselman  / IBM Research
Received on Friday, 22 March 2002 13:13:28 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 20 October 2015 10:46:09 UTC