Backing out the script changes in bug 343730 makes this SVG app work again.

*** Bug 359798 has been marked as a duplicate of this bug. ***

The main problem here is that nsSVGScriptElement doesn't implement DoneAddingChildren. However I also noticed that svg uses xlink:href to point to the url, not src as some of the new code assumes. I've got a patch in the works that takes care of this and some other things.

There we two major problems. 1. nsSVGScriptElement didn't override DoneAddingChildren so we never even got in to nsScriptElement::MaybeProcessScript 2. A couple of places assumed that the attribute name was src (rather than xlink:href for <svg:script>s) I fixed this and also improved the code that checks if a script element is "empty" (i.e. doesn't link to a script or contain inline script). This finally makes it possible to first give a script a text-child and then set that textchilds .nodeValue without the script executing too early.

Comment on attachment 244965 [details] [diff] [review] Patch to fix >Index: content/base/public/nsContentUtils.h >+ * NOTE! This method does not descend recursivly into elements. "recursively" >Index: content/svg/content/src/nsSVGScriptElement.cpp >+nsSVGScriptElement::HasScriptContent() >+{ >+ nsAutoString src; >+ mHref->GetAnimVal(src); Why use the animval instead of just HasAttr()

It's what GetScriptURI does. Not sure if it's possible to use animation elements to get an animated value without actually setting the attribute.

Comment on attachment 244965 [details] [diff] [review] Patch to fix Ah, ok. r+sr=bzbarsky with the spelling fix.

Comment on attachment 244965 [details] [diff] [review] Patch to fix > } May be here should be PRBool type? >+nsHTMLScriptElement::HasScriptContent() >+{ >+ return

> May be here should be PRBool type? > >+nsHTMLScriptElement::HasScriptContent() Same for nsSVGScriptElement.cpp

Checked in about a day ago with the return-types fixed.

Oh. Somehow I missed the mail about this getting fixed (seem to be doing that a lot). Thanks guys!