Question for net.views column in UNIX Today!

usinset!jr at uunet.UU.NET usinset!jr at uunet.UU.NET
Mon Apr 1 04:47:30 AEST 1991


Recently, you wrote:
> 
> 	*******************
> 	*   QUESTION #2	  *
> 	*******************
> 
> 	Is a single GUI standard really necessary?
> 
> 

Most definitely, yes!

A GUI standard defines a Look and Feel.  This has two parts to it.  The
first part is the `Look' definition; the second is the `Feel' guidelines.

The Look definition consists of a programmer's toolkit containing a
selection of widgets.  These widgets will be buttons, sliders, list
boxes, entry boxes, window size change knobs, etc, etc, which are likely
to be needed by many applications.  By using the same set of widgets,
all applications will appear similar on the screen, thus allowing the
user to instantly recognize parts of a new application, and hence the
user will immediately have an understanding of how to use the
application.

The second part, the Feel guidelines, is much more important.  All GUIs
embody a specification of how an application must behave in order to
conform. Guidelines include things like `the application should present
the user with a horizontal row of pull-down menu titles', and `if you
have a `File' menu, it must be the left-most pull-down', and `the `Quit'
button must be the bottom button in the left-most pull-down', and `a
menu is opened by clicking it with the left mouse button', and so on.
These kinds of `rules' ensure that applications behave consistently and
that users can navigate around them without having to reach for the
manual every time they need to do something.

The current proliferation of GUI standards is very annoying.  In my
organization we currently use X+uwm, X+MOTIF, X+OpenLook, MAC, Windows
3, and OS/2 PM.  Now, the technical gurus have no problem switching
between environments, although they do complain that having to remember
which mouse button to press is a bit of a pain.  However, non-techies
find the differences to be a major problem.  For example, a user used to
the MAC environment, finds using X+OpenLook very cumbersome.  The
differences in appearances of the widgets is not really a problem
because a push-button appearing as a rectangle or an oval is something
most people cope with automatically.  But, having to hunt around an
application to find its options, or having to learn a new set of
mouse/menu finger actions for each application is a bore.  I have
witnessed people asking other people to help them, simply because they
can no longer be bothered to learn how to use an application which uses
an unfamiliar style.

Compare this with another common user interface - that of an
automobile's driving console.  If you rent a car, you just open the
door, sit down and drive off.  You don't have to worry that the pedal on
the right is the accelerator, and on the left is the brake; you don't
spend 20 minutes with the manual to look for the turn signal knob!  Why
not?  Because things are in standard places.  This means that you can
drive a Honda Civic with as much ease as a Mercedes 500.  Sure, the
shape of the controls varies, and there are some common alternatives (2
pedals or 3, stick shift or column), but people can cope with this.

Having 6 different GUI Looks and many, many application Feels is too
much.  One standard, please.

A final note.  The single standard should not be rigid.  That is, it
should support customization of the interface, using something like a
`Preferences' dialog box.  It is important to allow users to set things
like color, mouse acceleration, button ordering, and so on.  An ideal
interface would also allow users to globally select details like where
in a window standard items are presented (i.e., scroll bar on right or
left, or maxmize button on title bar or in a menu), and to have this
take effect for all applications for that user.  Unfortunately, this
level of interaction with applications may be too complex for most
current GUI systems.

	Jim R Oldroyd
	Technical Director
	The Instruction Set
	Waltham,  MA 02154
	jr at inset.com


PS: Permission is granted to publish these comments in any form,
provided that the intent of the meaning of these comments is not
changed, and provided that an attribution to the author is included.



More information about the Comp.unix.misc mailing list