Lan Barnes wrote:
OK, so I'm supposed to drop everything K&R ever thought and rally around
the M$ interface guidelines?
I don't think so.
Ahem. Any standard that gives us the C syntax for declaring function
pointers has *zero* standing for any form of user interface debate.
And, if we're talking UNIX, VMS would like a word with you. There might
be a reason that VMS was the standard in universities for a long time;
they hated UNIX. UNIX got shoehorned in on the back of cheap workstations.
The reason for most of these GUI guidelines is that there is actual
research that shows that about .1 second of lag and people start to get
irritated. More than about 1 second without feedback and people change
context.
I find little reason to argue with that.
In addition, the whole "everything is a file you can pipe together" has
*also* been shown to be a particularly pernicious abstraction. Some
things work fine as files. The moment you add any form of metadata or
status in the channel, the UNIX way of piping things together breaks
down very badly.
It's a nice abstraction, and it has nice uses. You can pry a command
line from my cold, dead hands. However, there is a reason that I prefer
to script in just about *anything* other than sh, csh, bash, etc.
-a
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
