Re: bigger, sexier, more phly debian...
dlj0@lehigh.edu (David L. Johnson) wrote on 16.04.96 in <[🔎] 317448BB.386D78EB@lehigh.edu>:
> Stephen Early wrote:
>
> > There is no technical problem with creating a library with all the
> > commonly needed user interface elements in it. The problem is a political
> > one: persuading everybody to use it. Micro$oft managed it by introducing
> > a system that wasn't compatible with anything that already existed, and
... and that's similar to what I propose. Of course, you need to start out
with a system that already provides everything you want standardized,
otherwise it won't work.
> > constrained people to use the user interface elements that they had
> > created. Over the next few years, people gradually worked out ways of
> > putting their own user interface back into Windows (snazzy buttons, extra
> > types of list box, shadow/relief boxes, 'embossed' windows, etc.)
Now consider that just about nobody has the source to Windows/MacOS/
whatever. The source to this new thing, on the other hand, should be
available.
So it should be possible to do the snazzy stuff as an enhancement to the
base system, so that it would work for every program.
Similar things exist for the Mac, even without the OS source. Look at
"Aaron" for an excellent example.
That would keep what people like about X window managers, and yet still
keep similar style for all programs for more than window borders.
> Apple's approach was similar, and I understand that in order to write for
> the Mac you have to conform to their interface standards. But, I would not
> consider that a role model to emulate.
"You had to", in this context, means that you would be scolded, nothing
more.
I don't see what's bad with that role model. In fact, the results look
fairly compelling.
However, most of it was probably simply very good documentation.
> > Looks like people inventing their own user interfaces is a fact of life.
> > I don't like it, but I don't think there's much we can do about it other
> > than completely closing the development process, which won't work.
I don't think so.
> I don't really see the problem. Chaq'on a son gout.
The problem is that style decisions, with X, get made by program
developers, who usually are quite bad at it. Consequently, most X software
looks fairly shitty. And it also doesn't help that every program is
different.
There's two groups who should make those decisions: the designers of the
GUI system, and the end users. Program developers usually should stay out
of it.
My idea of a good basic toolkit would be one where a program using it
doesn't even know how most of the widgets are implemented, it just tells
the libs what functionality it needs. This, in turn, allows to change the
implementation in fairly drastic ways without touching the program.
All compatibility problems with new UI designs that I've heard of seem to
be from programs knowing too much about the UI innards.
Of course, a good object oriented language as a standard interface really
helps here.
MfG Kai
Reply to: