Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!cs.utexas.edu!yale!mintaka!wookumz.ai.mit.edu!rjc
From: rjc@wookumz.ai.mit.edu (Ray Cromwell)
Newsgroups: comp.sys.amiga.tech
Subject: Re: Help --> How to prevent the visit from the dreaded guru
Keywords: Flip answers are no help
Message-ID: <1991Jan6.032929.1813@mintaka.lcs.mit.edu>
Date: 6 Jan 91 03:29:29 GMT
References: <17130@cbmvax.commodore.com> <7465@sugar.hackercorp.com> <17133@cbmvax.commodore.com>
Sender: daemon@mintaka.lcs.mit.edu (Lucifer Maleficius)
Organization: None
Lines: 50


  On a side note Randell, how difficult would it be to add Resource
Tracking to the Amiga? Couldn't each library's close routine make
sure all resources are freed when a Task exits? I get really tired when
poor programs forget to free a Lock() or Memory. (XOper to the rescue)
It would also be nice to 'kill' a process and have it automaticly free itself.
Consider this:
  'KILL HandShake'
  Dos.Library's KillProcess() call does a sanity check, then proceeds
to close all the libraries said process had open. (I assume OpenLibrary
maintains a list and attacks it to the task) Each of the libraries/devices
free the resources from their close() routines. Poof! HandShake is
UnLoadSeg'd and all its resources have been cleaned up by the libraries
themselves.

 Virtual Memory shouldn't be a problem either. The OS merely checks for
the presence of an MMU and enables virtual mem (for FAST mem. Its impossible
to protect Chip mem from the blitter. And I can't imagine a machine
with an MMU that doesn't have fast ram)

MMU Protection. I'm fairly certain AmyDOS will never get Unix style
protection because of the shared-memory paradigm. But how about this
. AllocMem(blah,MEMF_PRIVATE)  Programs could request private
memory for their data structures and use Allocate/DeAllocate to
manage it. This would cut down on programs stomping on valuable data, so
a GURU could atleast exit with a warning 'Please save all data, then reboot'

I have no problems with Software Error requesters. I like them. I can save
my data/finish a download, then reboot. Or if developing, suspend the
task, recompile, and try again. GURUs aren't so forgiving. They
offer only 1 choice. Reboot immediately.
 
  Also. Alot of errors aren't fatal (Like FreeMem twice, etc) its the
0003/4 errors you have to watch out for (meaning a task when wild
stomping thru memory) I think FreeMemtwice type of errors should put
up a warning ('Task A went nutso, data may be corrupt, Wanna Reboot?')


Three more questions.

1) Has layers library been speeded up/reworked? It gets really slow
when there are lots of open windows on the screen.

2) What happens when you pull down a screen (on 2.0) and a productivety
mode screen is behind it? Can monitors actually handle scan rate changes
mid screen? (15khz <-> 31khz)

3) Was file notification/record locking ever added to the 2.0 FS?

Thanks in Advance. (I don't have 2.0)