Bug 6010

WebKit+SVG does not support proper SVGLoad events SVGLoad events are a bit tricky (and rather ill defined in the spec). In order to implement them properly, we need to do something like this: 1. When a close tag is hit on an SVGElementImpl, closeRenderer() (the current awful name) will be called. 2. At this point, we need to check to see if this element has an onLoad listener. 3. If this element has an onload listener, we need to check all of this element's children to see if any of them have externalResourcesRequired set (required resources), and if so, if those resources are not yet loaded. 3a. If all all children have already loaded their required resources, we issue an SVGLoad event at this time (with proper capture/bubble symantics). 3b. If NOT all of the children have loaded their required resources at this time, we must push some sort of data structure onto a "pending" stack (or toggle some flag on the element in question). 4. Once a load is completed for an already-closed element, we again have to check to see if all of it's required children have been loaded yet, and if their Load events have been sent. If all required children are loaded, we must then start at the deepest child who has not yet sent a Load event (who actually needs to, aka has a listener for one) and send that load event, then it's parents, etc. until we've reached the element who just finished loading. At which time we continue processing. One problem this may cause is possible DOM instability due to executing arbitrary code during parse time. We'll have to guard against this in some manner. *many* things in SVG are keyed off of this specific SVGLoad symantec. One simple example is Xlink as noted in 6009. Note, that after asking graouts and dino, we have recieved clarification from the w3c that SVGLoad events are *not* issued due to programatic insertion, only due to actual parsing.