> There are lots of things that could cause it to use 2.3Mb. Where
> do you register your event listeners? If you are registering
> individual event listeners on a large document (as per your
> example below it could take a bit).
Yes I register the mouse event like in my example...
> It sounds to me like something
> is just taking a long time when the document is loaded.
Then, what's the best method ?
> > I so think that it's not a problem in my code (maybe I'm
> wrong :-)), or
> > it's in a part that I don't know (this could be that :-))
> > (How do I trace... ?)
>
> Well on Windows if you run Batik from a DOS window you
> can press something like ctrl-'break' (the break key) and it
> will dump the stack trace for all threads. So you can look at what
> it is doing for those few seconds (i.e. who's code is running).
ok I will try. It won't be easy, I must be fast :-)
> It's probably worth doing a few times to get a more accurate picture.
>
> >>>Morever if I don't move the mouse over the canvas, updates made by
> >>
> >> > java (using updatemanager) are not showed. (This my main problem,
> >> > the delay is accessory)
> >>
> >> So this really sounds like you are not using the UpdateManager's
> >>RunnableQueue. Can you post some code that shows how you post the
> >>runnable? Also an example runnable would be useful (best of course
> >>is a standalone reproducable test case).
> >
> > Well in fact I'm wrong, the updates are done, without moving on the
> > canvas.
>
> Ahh, that is good to hear.
:-)
> The code below looks good to me.
Yes, of course, it comes from the mailing list :-)
Before 1.5 final, I haven't used updatemanager, but in the 1.5, without
it, nothing worked for me, no updates.
>
> > For the runnable :
> > queueAction.invokeLater(
> > new Runnable() {
> > public void run() {
> > elt.setAttributeNS(null, "visibility", "visible");
> > }
> > }
> > );
>
>
>
> > where queueAction =
> canvas.getUpdateManager().getUpdateRunnableQueue();
> > This is not the method I use to highlight a shape but only the
> > setAttributeNS part changes :
> > elt.setAttributeNS(null, "class", "highlighted");
>
> Hmm, I recently fixed a number of performance and propigation
> bugs in changing 'properties' in CSS. I am skeptical that
> the performance issues effect you since they would effect you
> for all changes not just the first one, but it is probably worth
> while to upgrade to current CVS if you are doing things like
> the above.
Well I've haven't access to the CVS port because of proxy... Is there
another mean... ? (furthermore I prefer the binaries)
> > OnAction onAction = new OnAction(dynUpdt, this);
> >
> > Element gSets = svgDoc.getElementById("sets");
> > NodeList nodeList = gSets.getChildNodes();
>
> Couple of points here. First off DOM Event 'bubble'
> through the DOM tree so if you just registered listeners
> on the gSets element you would recieve any events that
> happen on the child elements.
Well euh... because of not being english, I'm not sure to understand
this sentence :-} (the beginning).
Are telling me that by doing this, it could works ? :
. register my onAction object on gSets
. When onouseover event occured, the event object is for example a rect
in the g parent element ? Not the g element itself ?
>
> > int i=0;
> > Node elt;
> > while ((elt = nodeList.item(i++)) != null) {
> > if (elt.getNodeName() != "#text") {
>
> A better check is: (elt.getNodeType() == Node.ELEMENT_NODE)
> This faster and means you won't get tripped by CData sections or
> Comments.
Ok thanks, indeed I found my method, too... too... dirty :-)
>
> > EventTarget t = (EventTarget)elt;
> > t.addEventListener("mouseover", onAction.new Over(), false);
> > t.addEventListener("mouseout", onAction.new Out(), false);
> > t.addEventListener("click", onAction.new Click(), false);
Morever a better method would have been to have one of three objects in
onAction, to avoid creating one on every element...
Thanks
Olivier
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
