20491 2012-12-22 00:45:53 +0000 [Custom]: Need to actually define the upgrade algorithm so it's interoperably implementable 2013-04-10 20:49:20 +0000 1 1 1 Unclassified WebAppsWG HISTORICAL - Component Model unspecified PC All RESOLVED FIXED P2 normal --- 17103 1 bzbarsky dglazkov public-webapps-bugzilla oldest_to_newest 80519 0 bzbarsky 2012-12-22 00:45:53 +0000 Spec says: > "For each element ELEMENT in TREE whose custom element name is NAME: " In what order? What happens when the various events that fire during this algorithm trigger handlers that mutate TREE? This thing needs to be very clearly defined. As a start, I would recommend not firing any events until after all the other work is done. 81291 1 dglazkov 2013-01-12 00:38:14 +0000 (In reply to comment #0) > Spec says: > > > "For each element ELEMENT in TREE whose custom element name is NAME: " > > In what order? Good point, needs to be in tree order. What happens when the various events that fire during this > algorithm trigger handlers that mutate TREE? Can you help me understand what you mean here? The algorithm does suppress observers currently (although this might not be the right thing). Which other events are you talking about it? > > This thing needs to be very clearly defined. As a start, I would recommend > not firing any events until after all the other work is done. If you're talking about elementreplace and elementupgrade events, I don't see any harm in them being fired after each upgrade? 81293 2 bzbarsky 2013-01-12 01:08:18 +0000 > Which other events are you talking about it? Mutation events, for one thing. But also, suppressing mutation observers may well not be ok. See the separate bug filed on that. But most importantly, the elementreplace and elementupgrade events. > I don't see any harm in them being fired after each upgrade? Let me re-ask: What happens when the various events that fire during this algorithm trigger handlers that mutate TREE? 85883 3 dglazkov 2013-04-10 20:49:20 +0000 The spec has been updated to address this a while back.