Hi CHARLES! :-)

 CA> The 'huge' benefits to everything being just a file are more a
 CA> benefit to programmers unwilling/unable to write automated
 CA> detection of file types with proper error trapping to avoid
 CA> foul-ups.
This can be done even under the Unix model, all the programmer needs to do
is use one stat() syscall and look at the resulting information.

The model of other operating systems, where files and devices have
different ways to access them leads to more code in programs and thus more
bugs. Also, there are often different interfaces to, say, serial ports and
printer ports. And thus each kind of device gets their own interface (the
Windows way) and the programmer needs to learn all of them - and they all
work differently, which increases the risk of the programmer making a
mistake. Under Unix, it's open(), then write() and/or read(), to any
device. Always works the same way.

 CA> The days of all systems persons having programming
 CA> experience are in the past. 
Still, some knowledge of the system is required. I have often seen Windows
users tweak settings in their systems config to and fro without knowing
what they do. The reason being "I have seen it working for someone
when they did twiddle this setting". This may be no problem on a home
system, but even those are nowadays often connected to the internet. I
wouldn't say programming experience is needed, but the basic principles of
the system one needs to know.

 CA> I would advocate alternatives to 'cp' that are restricted
 CA> within the binary itself rather than rely on file attributes
 CA> (permissions).
Sometimes the admin needs to be able to cp to a device file (or more likely
dd). Also, default permissions on hard disk devices and such are to deny
normal users write access.

Root is allowed to do everything, but that is sometimes necessary to repair
a damaged system.

What you advocate could be done, it's just that nobody so far seems to have
found it necessary because basically normal user accounts are safe to use,
no matter what you happen to type in. All that can happen is that a user
accidentally deletes all his own files.

 CA> I am convinced that daily use by less
 CA> knowledgeable persons as a desktop environment will only be
 CA> practical when the average user cannot trash the system even if
 CA> they try to. 
You can't trash the system from a user account. Only root can do that - and
all distributions today setup a normal user during installation. In
addition, root gets a set of tools (GUI stuff) to manage system
configuration, thus reducing the risk of the admin making silly mistakes.

Trashing the system is all relative, of course. No system is secure without
backups - hard disks can go bad, or even the whole system in case of a
lightning strike.

Back when I used DOS and Windows, I almost never made backups and as a
result have lost many things that I wrote in those days. I switched to
Linux in 1998, and I still have all the documents and programs from then
that were worth keeping.

The Unix way gives all power to the admin, including the power to shoot
your own foot or blow away your whole leg. With power comes responsibilty.
;)

Ciao
Pascal

--- Msged/LNX 6.1.1