For performance reasons, Vladan recommends we start Firefox Health Report after session restore, not final-ui-startup. This patch accomplishes this. This will influence our metrics for session active/idle time since we may miss observers fired between app startup and session restore. We had that problem already. This just makes it worse. When we rewrite session code in bug 827157, we should come up with a more robust solution.

Comment on attachment 699523 [details] [diff] [review] Change observer notification, v1 Review of attachment 699523 [details] [diff] [review]: ----------------------------------------------------------------- Seems formulaic. Is sessionstore-windows-restored sent when you opt *not* to restore a previous session? https://developer.mozilla.org/en-US/docs/Observer_Notifications is ambiguous. r+ conditional on that being true.

If there is any interest, I can expose a promise that is resolved once session restore is complete.

(In reply to David Rajchenbach Teller [:Yoric] from comment #2) > If there is any interest, I can expose a promise that is resolved once > session restore is complete. That would be interesting. But, we need something today that works. Is or is not the "sessionstore-windows-restored" event always fired?

(In reply to Gregory Szorc [:gps] from comment #3) > That would be interesting. But, we need something today that works. Is or is > not the "sessionstore-windows-restored" event always fired? It's always fired, we rely on it in Telemetry ping http://mxr.mozilla.org/mozilla-central/source/toolkit/components/telemetry/TelemetryPing.js#1058

Comment on attachment 699523 [details] [diff] [review] Change observer notification, v1 [Approval Request Comment] See comment in bug 804745.

Comment on attachment 699523 [details] [diff] [review] Change observer notification, v1 https://bugzilla.mozilla.org/show_bug.cgi?id=804745#c37

Pushed to Aurora: http://hg.mozilla.org/releases/mozilla-aurora/rev/50db9b596a5d