#Multi-user Amiga?
7 messages in this thread
Don:
I keep reading this thread, and am thoroughly confused.
Our VAXen at work occasionally crash. The whole multi-user, multi-processing
system goes on vacation. Surely, this is a "true" multi-user machine.
When ever part of the system gets changed, our vulnerability to such crashing
goes up. After a while, as bugs are found, work arounds implemented, that
vulnerability goes back down to very low levels, so that we mainly need to fear
hardware, rather than software, problems.
That, of course, means that our DP guys DO NOT permit the rest of us free run
of the system. We must "run their applications" for the system to be secure.
They keep the users for the most part on a tight leash.
Granted, the Amiga is not currently used much as a multi-user machine. Indeed,
the OS was not designed with that in mind. But, even on the VAX, where both of
these things are part of its raison d'etre, I doubt you'd find many "muti-user"
systems where the DP managers did not very carefully control what users could
and could not do on the system. So, why is Thol's doing this on his
hypothetical Amiga multi-user system somehow "cheating".
Ray Lewis
Ray,
I do the exact same thing on my UNIX boxes at work. OF our 3,000
users…only about 10 can do anything except access the database thru a very
controlled environment that I set up.
However, those 10 have pretty much free rein (other than superuser
access) and generaly are doing developement or bug fixes on current
applications.
Should one of them generate the equivalent of a GURU (Unix core dump)
no one else has any notion that there was a "crash"…since the multi-user OS
provided user-user protection.
The same can't be said for the Amiga. Should any one user generate a
crash…the whole machine is down. Tom's application seems to prevent an
intentional crash by one of hte users…but what about the unintentional?
Tom's not cheating…I never said he was.
Don
If a mainframe database application with multiple users drops dead due to a bug
in the database which one of the users happened to trigger, it's highly likely
that all of the other users will drop dead, too. Is that an example of one
user crashing a system? Is the operating system at fault? Is the system
manager at fault, for using an installed image (so that all users were sharing
the same code in memory) rather than duplicating the code all over RAM?
A controlled-access Amiga application will prevent any user from crashing the
system intentionally or unintentionally, other than by performing a legal
operation in the database which causes the database to halt; and this problem
would be most likely detected by the users' front end as a database timeout,
and allow an orderly exit.
Tom,
Users of the database will crash, the database itself may crash, but
the machine (system) won't crash.
Don
So if I have GOMF in the Amiga (or AmigaDOS 1.4?) to prevent a program bug from
halting other tasks (or if I tell the system operator not to acknowledge
software failure requesters), the machine itself will not crash.
The problem is, GOMF can be over-written by a user program! The amiga won't
work really well as multi-user 'till it has hardware memory protection built
into the OS.
-MIke
Tom,
I agree with what Mike said…and would add that from the comments I've
seen here…GOMF has problems all by itself. That is NOT from personal
experience…I don't use GOMF.
Don