Re: autocompletion

<quote who="Ricardo Fern?ndez Pascual">

> I didn't mean to despise Havoc. I am sorry if it sounded like it. I
> actually agree with most things he says and with the whole idea of
> simplifying interfaces in general and preference dialogs in particular.
> 
> But in the same thread you had mentioned the famous Havoc document about
> that subject. I'm a bit tired of seeing it *every* time that someone even
> thinks about adding a pref.

I raise it when someone asks why the discussion about adding a preference is
important, "Why can't we just do it?" In this case, it was very specifically
asked for. It's an important piece of documentation to understand how we're
tackling usability in GNOME, and I feel that it's balanced and accurate. If
it were a document that said "all user-visible preferences are bad", then I
would not be interested in using it as an educational device. :-)

> In some cases there's no "right" way to do something[1]. I think this is
> one of those cases[2]. 

There's two problem areas for 'the right way': behavioural and preferential.
I'm fairly certain that this discussion is mostly about behaviour - what
works, what doesn't, consistency, interrupting the user, accessibility, etc.

More often than not, you can find a good balance between these. Sometimes,
you can't... Your choice then is to either trot out another preference, or
think hard about your audience and play for the greatest common factor
(*not* the lowest common denominator - that's not what usability is about).

Yes, it does take a significant amount of effort to do this. Hackers, more
often than not, will find it easier to just throw more code at the problem,
because that is much less of a hassle. That's not a bad thing, it's just not
going to solve the usability problem fast. :-)

I'm pretty sure this can be solved without resorting to even a GConf-only
pref. If people are really passionately disgusted with the drop-down, you
gotta work out why... There's no point giving in because people give you
shit rather than useful input.

> [1] There are more cases. For example, bug #86108 (which i think is not
> going to be fixed anytime and makes imposible for me to use metacity).

[ Sure, apart from the complex issues with focus, DND, etc., I generally
agree with the concept that sloppy focus shouldn't raise windows with
interaction. I haven't missed it much whilst I've been using metacity, but
if the behaviour was *just right* when it returned, I'd welcome it with open
arms.

But that's the clincher - the behaviour should be "correct". When someone
fronts up the code, we'll all be better for it. I don't think Havoc is being
unreasonable in this instance - he's sticking to his guns to get it right. ]

> [2] I have thought about this a lot because autocompletion is my pet
> feature. What some people find terribly useful annoys other people a lot.

Let's get some usage cases and solid information; I'm happy to help.

- Jeff

-- 
    "It's not sufficient to 'use simple words to explain things'. Things    
      must actually *be* simple, which is much harder." - Martin Pool       


-------------------------------------------------------
This SF.net email is sponsored by: Get the new Palm Tungsten T 
handheld. Power & Color in a compact size! 
http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0002en