Path: utzoo!utgpu!water!watmath!clyde!att!ihnp4!cbmvax!uunet!pyrdc!netxcom!ewiles
From: ewiles@netxcom.UUCP (Edwin Wiles)
Newsgroups: comp.sys.amiga.tech
Subject: Re: Suggestion for 1.4
Message-ID: <858@netxcom.UUCP>
Date: 12 Jun 88 00:43:40 GMT
References: <8806021714.AA12190@jade.berkeley.edu> <611@myrias.UUCP> <18191@cornell.UUCP>
Reply-To: ewiles@netxcom.UUCP (Edwin Wiles)
Organization: NetExpress Communications, Inc., Vienna, VA
Lines: 33

In article <18191@cornell.UUCP> blandy@crnlcs (Jim Blandy) writes:
>In article <611@myrias.UUCP> cg@suncg.UUCP (Chris Gray) writes:
>>And now the blue sky: I've heard a weak rumor that 1.4 will be able to use the
>>68851 MMU to provide protected address spaces. Any form of this, even if it's
>>not the default, would be great.     [ stuff ]
>
>Do we REALLY want protected address spaces?
>
>It seems to me that a lot of the most interesting ideas out here
>(ARexx, something about Message Brokers, etc) involve fiddling with
>other people's address space.
...
>I strongly agree that one haywire program shouldn't crash the system.
>You should be able to crash as much as you like, in any way you like,
>without disturbing other tasks (well, I don't LIKE crashing... :-).
...
>Or is there a neat way to get around these problems (wouldn't that be
>superb) that I'm missing?

I think so.  The only reason for crashing a system is if you cannot guarantee
that the operating system, and its associated *critical* information, is
uncontaminated.  Thus, the only part of memory that must be protected, is
that which is used by the OS.  (Anything else is on it's own.)  Even better
would be a mechanism where you can tell the OS/MMU to protect a program and
its data, or can tell it to leave it open.

Is this possible?

-- 
...!hadron\   "Who?... Me?... WHAT opinions?!?" | Edwin Wiles
  ...!sundc\   Schedule: (n.) An ever changing	| NetExpress Comm., Inc.
   ...!pyrdc\			  nightmare.	| 1953 Gallows Rd. Suite 300
    ...!uunet!netxcom!ewiles			| Vienna, VA 22180