smaug commented on bug 269132: "How do you handle the case where someone is modifying the subtree of the repeat element dynamically? Shouldn't you listen DOMSubtreeModified and on event call Refresh() or something like that." He's probably right :)

Testcase. XTF is not too happy about it: ###!!! ASSERTION: element not implementing nsIContent!?: 'content', file nsXTFFrameUtils.cpp, line 51 Break: at file nsXTFFrameUtils.cpp, line 51 WARNING: NS_ENSURE_TRUE(aContent) failed, file nsFrameManager.cpp, line 372

Alex, do you have a comment on the assertion message?

(In reply to comment #2) > Alex, do you have a comment on the assertion message? Olli had a clue for it. We returned a nsnull insertion point...

Here's a rough patch, should work for non-nested repeats, but I have not tested it much.

(In reply to comment #4) > Created an attachment (id=180795) [edit] > Rough patch > Does this work also when you modify something, which is not a child of the repeat element, but further descendant?

(In reply to comment #5) > (In reply to comment #4) > > Created an attachment (id=180795) [edit] [edit] > > Rough patch > > Does this work also when you modify something, which is not a > child of the repeat element, but further descendant? Nope. I should possibly have mentioned that *ahem*. So admittedly, it's a sorry excuse for a fix, but then we're back at the mutation events, aren't we? And "somebody" keeps telling me to avoid using that for now ... or something :)

(In reply to comment #6) > (In reply to comment #5) > > (In reply to comment #4) > > > Created an attachment (id=180795) [edit] [edit] [edit] > > > Rough patch > > > > Does this work also when you modify something, which is not a > > child of the repeat element, but further descendant? > > Nope. I should possibly have mentioned that *ahem*. So admittedly, it's a sorry > excuse for a fix, but then we're back at the mutation events, aren't we? And > "somebody" keeps telling me to avoid using that for now ... or something :) I've looked a bit into the mutation events, and they are rather expensive aren't they? I'm tempted to let us support what I have here, but WONTFIX anything more than that... what do you say? Dynamics (in general) is not a must....

RIP xforms