GNOME Bugzilla – Bug 649566
Incorrect uses of the gtkSwitch widget
Last modified: 2011-08-26 17:35:54 UTC
From bug #644657 and the discussions that happened around it, my understanding of the acceptable use of a gtkSwitch widget is for things that toggle hardware on/off, and it is not acceptable for simply toggling boolean settings, which would be better served by a checkbox widget. Here is a proposed list of areas in gnome-control-center 3.0 where the gtkSwitch should be replaced by a checkbox: - Screen lock: make the "Lock screen after:" label a checkbox label - Bluetooth: the "visibility" setting (at least) - User account: auto login - Date and time: 24 hours mode - System info: forced fallback mode You may find others but I'm taking a reasonable/modest approach here instead of saying they should all be taken out indiscriminately :)
(In reply to comment #0) > From bug #644657 and the discussions that happened around it, my understanding > of the acceptable use of a gtkSwitch widget is for things that toggle hardware > on/off, and it is not acceptable for simply toggling boolean settings, which > would be better served by a checkbox widget. Hardware is one key case; so too are toggling specific 'services' or modes which modify the system in a major way [1]. > Here is a proposed list of areas in gnome-control-center 3.0 where the > gtkSwitch should be replaced by a checkbox: > - Screen lock: make the "Lock screen after:" label a checkbox label > - Bluetooth: the "visibility" setting (at least) > - User account: auto login > - Date and time: 24 hours mode > - System info: forced fallback mode > > You may find others but I'm taking a reasonable/modest approach here instead of > saying they should all be taken out indiscriminately :) I agree about 24 hours mode (see bug 640059); the others seem to be within the guidelines. Visibility might be on the border, I guess... [1] http://live.gnome.org/Design/Whiteboards/SwitchGuidance
My criterion on which I based many of those things was that most of these things do not an immediately visible effect and only apply on the next login. As such, this "on/off" switch metaphor from the real world breaks down completely. It's a preference, not something that flips electric current flow.
(In reply to comment #2) > It's a preference, not something that flips electric current flow. But I think that is the mental model we're alluding to. Auto login might not come into effect until the next time you login, for example, but the suggestion is that it is engaged from the moment you flip the switch. That's easier for people to understand.
As per Allan's comment, we believe that those uses are within the guidelines. The 24h mode got fixed in a separate bug. (And the Visibility switch in Bluetooth does actually have "electrical" and instantaneous effects, fwiw).