Martin Nordholts ([EMAIL PROTECTED]) wrote:
> >> If you agree on that, then this improves consistence.
> >
> >I don't agree. First, determining what the "most common use" of a button
> >is, probably is subjective at all. And even if we could determine this
> >in an unambigous way we'd end up basically with a list: This button
> >shows a dialog, this button does not. Shift toggles this. The user would
> >have to learn the behaviour of all the buttons by heart, since there is
> >no simple rule to describe the behaviour all over the application.
> I strongly doubt that any regular user had any idea that that was the 
> logical connection between a Shift and a non-Shift click. I had no idea.

I did state explicitely that I don't claim that GIMP is consistent at the
moment. There is no doubt that the current situation can be improved. It
however IMHO cannot be improved by randomly changing the behaviour of
individual buttons.

> As I said, I always figured GIMP wants to be _effective_ to work with 
> rather than logical. I mean there are a lot of inconsistences in the UI, 
> however, they make it _effective_ to work with GIMP.

So to use GIMP effectively you have to go through a list of dozends of
buttons and memorize what they do when clicking and what they do when

My idea of consistency is: The user looks at a button and can predict
(maybe after learning one or two rules) what the button will do when it
gets clicked, and what it will do when it gets shift-clicked. Even when
he encounters this particular button for the first time.

Your approach involves a lot of guesswork: We have to make an educated
guess, what is the more frequent usage of each of the dozends of
buttons. Each user has to guess what we guessed. And I bet that there
are users out there who don't agree with our guesses. How do we defend
our guesses?

To exaggerate your approach: Wouldn't it be the most usable GIMP if we
keep counters for the button presses? And if it turns out that a user
uses SHIFT-Click more than a simple click, shouldn't we just switch the
behaviour of the button on the fly? Maybe we should also reorder the
buttons according to the button click count.

According to your arguments this would result in the most usable GIMP
ever, because we react to the individual usage pattern of each user. But
it basically will result in a Chaos, where it is unpredictable what a
specific button will do, even where it will be located. It is impossible
to document.

I firmly believe that simple consistency rules are important for
usability. And that saving 0.3 seconds per new-layer (assuming for 0.5
seconds, that your made-up number resembles reality) or 30 seconds per
100 new layers is not as important as a goal.

Conclusion: We definitely can improve the GIMP, but we need a
coordinated approach. A usability study would help here.


              [EMAIL PROTECTED]    

Gimp-developer mailing list

Reply via email to