I think you didn't understood, so imagine this scenario (auto-hide on):
1. You move the mouse away from AWN (so awn_hide() is called and auto-hide timeout is started).
2. While AWN is still visible, you change your mind and want to launch another app, so you move your mouse back on AWN.
Current behaviour: AWN auto-hides once the timeout expires, no matter that you have your mouse back on it.
Expected behaviour: Timeout is cancelled, and AWN is not auto-hidden while you don't move your mouse away from it again.
And btw the patch was written against trunk R35, so I did see the 2 second timeout.
I think you didn't understood, so imagine this scenario (auto-hide on):
1. You move the mouse away from AWN (so awn_hide() is called and auto-hide timeout is started).
2. While AWN is still visible, you change your mind and want to launch another app, so you move your mouse back on AWN.
Current behaviour: AWN auto-hides once the timeout expires, no matter that you have your mouse back on it.
Expected behaviour: Timeout is cancelled, and AWN is not auto-hidden while you don't move your mouse away from it again.
And btw the patch was written against trunk R35, so I did see the 2 second timeout.