GNOME Bugzilla – Bug 591084
Usability issue when face browser is turned off
Last modified: 2010-05-03 21:08:18 UTC
When the GDM login dialog starts up, you need to click on the "Log In" button to start the process of entering the username and password. If you fail to authenticate by typing a wrong password, it takes you back to the original window where you need to click on the "Log In" button to start the process over again. Needing to click on the "Log In" button each time you mistype a password is annoying. It would be nicer, I think, if it just went back to the username entry on failed login.
Why is there a "Log In" button at all? The user has no other meaningful options available to them at that point in the login process, so fundamentally we are just forcing them to press a button for no good reason. Surely we should we just dispense with the "Log In" button entirely, taking the user directly to the field where they can enter their login credentials?
Starting the PAM conversation when the user isn't around could have bad consequences. For instance, it would mean fingerprint scanners would get turned on and would stay on until the user gets back (which some scanners aren't designed to do).
I don't mind having to hit the button to start the PAM conversation. But it is annoying to have to press the button over-and-over on each failed login. It seems that it would be nicer and more usable if the PAM conversation were just restarted after a failed login. If there is an issue with certain PAM stacks not liking to be left in this state, then it would be reasonable for the login GUI to timeout after a minute or two and revert back to the initial screen if the user stops trying to login.
Thanks to Rays comment I now understand why the "Log In" button was added; it makes sense in the light of devices like fingerprint readers, which I hadn't considered. However, I still think it makes the case for the most common input device (the keyboard) sub-optimal. And I now wonder if we can enhance Brians idea, and use a timeout to both take into account Rays hardware issues, and yet still satisfy people like Brian and I who don't want to have to press a button before entering our credentials. So, we could go straight into the PAM conversation without needing to press a button to start. That conversation would consist (as now) of a loop of interactions with the user (where each interaction is the gathering of credentials from the user) until we have valid credentials. At the start of each interaction we would set a time-out so that if there is no input within a specific time, then we can break out and switch off any hardware (like fingerprint readers that cannot be left running indefinitely), present the "Log In" button, and wait for the button press before re-enabling any necessary hardware and restarting the overall PAM conversation again. This would seem to allow us to avoid any extraneous button presses, and still allow for correct power-management of devices like fingerprint readers. Thoughts?
sounds reasonable.
What I do not understand, besides having to click a login button or not, is why gdm seems to accept keyboard input anyway, even when "login" was not clicked. It does not log you in, but there appears a text entry field when the user starts typing right away. The entered text seems to go just nowhere. This might even constitute a slight security risk, when a user starts - blindly - typing his username and password right away without clicking login, because the password will then be visible on the screen in clear.
This is addressed by fixing bug #591082. *** This bug has been marked as a duplicate of bug 591082 ***