Instant apply and explicit apply in KDE HIG

[Gábor Lehel]

On 3/3/06, Ellen Reitmayr <ellen at kde.org> wrote:
>
> On OpenUsability, Stefan Monov suggested to promote the usage of instant apply
> instead of explicit apply in certain situations in KDE
>
> see:
> http://openusability.org/forum/message.php?msg_id=1991
> http://developer.gnome.org/projects/gup/hig/1.0/windows.html#instant-apply

I like instant apply a lot. Apart from being more convenient (you
don't have to separately click the apply button -- but this is the
lesser advantage), it also offers instant feedback, so the user can
usually tell what the option (s)he just clicked actually changed (and
then change it further or back as a result), and it's also more
lifelike -- in real life, when you change things, they change, and
don't ask you to confirm first.

> I replied that while instant apply is a useful thing in certain situations,
> there undoing operations becomes problematic:
>
> <snip>
> for example, in kivio, there is instant apply in the docker windows.
> when i modify the x location via the geometry docker, i can either enter a
> value manually (no prob here), or i use the spin box buttons. if i use them
> and i want to revert my changes, i have to press 'undo' as often as i pressed
> the button in the spin box. that is crazy!
> </snip>
>
> He proposed to remember only the changes that change direction:
>
> <snip>
> A possible approach is to remember every change that changes the direction.
> Following your example with the spinbox:
>
> in the beginning - 1
> changes to 2
> ...3
> 4
> 5
> 6
> 5 (this change saves in history)
> 4
> 3
> 4 (this one too)
>
> This would also be consistent with the current editbox undo/redo behavior: the
> last solid sequence of backspaces/characters (whichever was last done).
> </snip>

Well, clicking on spinbox arrows seems like a bit of a special case to
me. Certainly you don't have this problem when changing checkboxes,
radio buttons, combo boxes, text fields (which have their own undo
functionality), and so on. (Nor if you hold down the spinbox arrow,
instead of clicking on it repeatedly.)
The proposed solution also seems alright to me. Or just using a timer
-- if the user hasn't touched it for N seconds, start a new entry in
the undo history.

>
> I'm tempted to think it's a good idea to promote instant apply in certain,
> defined situations - assuming that the history works.

Personally, I'd like if instant apply were promoted *except* in
certain, defined situations (basically, the ones the GNOME HIG you
linked mentions). If it's a positive change, why be conservative with
it? (And if it's not, why promote it at all?)

>
> What is your opinion??
> And: Won't it be difficult for users if instant and explicit apply are mixed?

Well, they already are mixed to a degree (opacity sliders, volume
controls, menu options, etc. are instant apply), but yeah, consistency
is key. I don't think a select few dialogs having an Apply button
would cause too much confusion (but then, I'm not an oracle or
anything).


--
Work is punishment for failing to procrastinate effectively.