The frameloader switch added in bug 113934 sends out only very sparse notifications that things have changed. You pretty much have to listen for pageshow if you want to know something has happened. We should try to make this send out more common notifications that apps/extensions may be expected and doing things based on.

This is a testcase that checks that certain sequences of progress and event notifications are sent by some different types of loads for a browser. It isn't a terribly smart testcase, it will timeout if events it expects don't happen. Currently all but test 6 and 7 pass. It currently only sees a pagehide and pageshow event. The other notifications are what I think should be sent out with the docshell switch. I think we might also want to send some history listener events of some kind, given that the switch totally changes the history list for the browser, but I'm not sure what would be appropriate short of an OnHistoryPurge and then OnHistoryNewEntry for each url in history. I also wanted to include testing for DOMTitleChanged which I think we should think about dispatching for the docshell switch case, however that is dispatched async so it becomes more difficult to verify its presence.

Dave: is this gonna be late-compat? If so, can you mark the bug and cc: fligtar and blizzard so they know for add-on developer outreach?

If it happens then yes

This doesn't look like it's going to happen for 1.9.1, so I think adding late-compat is a bit premature. If that's wrong feel free to add it back...

Oh, sorry, was going through old bugmail and saw it had been removed during a change whose only other effect was add a CC, so I jumped to the obvious conclusion. Probably should have skimmed the bug first to be sure, I guess, although I'd bet 99% of the time such a change is unintentional.