Instant apply and explicit apply in KDE HIG

[Cornelius Schumacher]

On Monday 06 March 2006 11:01, Ellen Reitmayr wrote:
>
> Well, I did not mean that instant apply would not work in most config
> dialogs, but it's benefits do not outweigh the costs. the costs are a high
> effort to change existing dialogs, and a long conversion phase where each
> application has different mixes of explicit and instant apply dialogs (some
> will redesign all the config dialogs, some only fonts, some will only
> change the newly added dialogs).

Of course we would have to try hard to keep consistency and make it clear 
where instant apply is used and where not.

Configuration dialogs might in fact be not so difficult from the 
implementation side. With the wider adoption of KConfig XT and concepts like 
KConfigDialog we might even have a central place to change the behavior of 
many dialogs.

> While most people on this list and other power users will have few problems
> to handle such inconsistencies, it will be hard for less experienced users.
> If we limit the scope of instant apply to some, defined situations,
> possibly in combination with a different UI approach such as dockers, it is
> easier to care for a consistent implementation of this new approach.

This sounds like a good approach. If the dialogs (or dockers or whatever) that 
use instant apply are clearly distinguishable from conventional config 
dialogs, it will certainly help to avoid inconsistency feeling for the user.

Do you have any concrete ideas how this could look like? If I for example look 
at the config dialog of KOrganizer I think that almost all options would work 
better or at least as well as before with instant apply. How could I change 
this without confusing the user?

> I don't want to seem conservative, but I want to avoid situations where an
> apparent 'usability improvement' leads to a mess of inconsistent
> implementations - and in the end makes KDE less usable, especially for
> non-power users.

As instant apply can be so much better than our current approach I think it's 
worth thinking about how we could do the big step of getting it wide adoption 
in KDE.

-- 
Cornelius Schumacher <schumacher at kde.org>