Comment 11 for bug 944161

My magical test script doesn't shed any light on the matter. But I have been looking over Orca's debug.out output -- further enhanced with even more info. The order of events is this:

1. Orca is started up.

2. Orca goes through its initialization which includes starting up speech, braille, registering event listeners, registering keyboard listeners, looking for the active window, presenting the active window to the user (in this case that would be the frame named 'Login Screen'), and finally it starts up the registry.

3. The first accessible event we get in Orca upon starting the registry is for the first key press. In other words, Orca is not receiving a 'hey, the password text object has just claimed focus!'

The reason that arrowing down and up makes the problem go away is because Orca gets focus-related events at that point. Because unity-greeter has told Orca focus is in a password field, it does not echo keystrokes and share your password with anyone in earshot.

As for why the events are not showing up in Orca but are in my test script, I'm not certain:

* Maybe timing?? My test script does absolutely nothing but start up and print stuff out. Orca's doing a lot more than that.

* Perhaps instead it is a side effect of how Orca is getting started up??

Getting to the bottom of that, however, is key.