Comments on: What should the action name for an indeterminate checkbox be?

By: Marco

Marco — Fri, 13 Feb 2009 13:35:31 +0000

@Michael and @Rich: This is not about states (the current state), but the action that performs a change. On most elements, this is the very general action of “click”, and for Firefox/Gecko in general, a checked checkbox exposes an action name of “uncheck”, and an unchecked checkbox exposes an action name of “check”. But since we do not know what the next state of a currently undetermined checkbox will be, the action name must be something meaningful, but general enough so that the next state can be either checked fully, or unchecked. From other comments in the bug and what I got via e-mail, it looks like we’ll call this one “Cycle”.

By: Michael Marshall

Michael Marshall — Fri, 13 Feb 2009 12:01:06 +0000

i agree with comment number 2 it should say partially checked

By: Wouter

Wouter — Wed, 11 Feb 2009 19:08:52 +0000

In GTK+ this is called “inconsistent”, e.g. gtk_check_menu_item_set_inconsistent()

By: Rich

Rich — Wed, 11 Feb 2009 16:41:09 +0000

I think toggle is misleading here as well. I’d suggest “partially checked” (this is what Jaws says when it runs into a similar windows control).

By: Steve

Steve — Wed, 11 Feb 2009 14:40:46 +0000

Currently in AT-SPI there is a discrepancy. The user has a single gesture to change to the opposite state depending on the current state but the actions define in specific end states (check or unckeck). Is there a good reason for this? If the AT wants to give the user the same experience as the direct user then it must look at the current state and use the correct action to give a similar toggle effect.

Following that logic you should expose both Check and UnCheck in the indeterminate state and let the AT do all the work. Yuk – how did we get here?

If there is a good use case for an action that ends in an unknown state as determined by current state and application logic then Toggle might do as you suggest. To my mind toggle indicates change to the other state of 2 states so isn’t right for tri state logic where Cycle, Change, Alter. NextState, might be better suited

Hey how about