peter@sugar.UUCP (Peter da Silva) (06/13/88)
The canonical answer: make all memory not allocated MEMF_PUBLIC read-only to all other tasks. The canonical problem: people using statically allocated data for stuff that other tasks have to write (for example, FileInfoBlocks). The solution: how about trying it and seeing how much stuff breaks. -- -- Todays stupid pet trick: replying to an IntuiMsg after you've already -- closed the window. -- -- Peter da Silva `-_-' ...!hoptoad!academ!uhnix1!sugar!peter -- "Have you hugged your U wolf today?" ...!bellcore!tness1!sugar!peter -- Disclaimer: These may be the official opinions of Hackercorp.
jimm@amiga.UUCP (Jim Mackraz) (06/16/88)
In article <5580@xanth.cs.odu.edu> kent@xanth.UUCP (Kent Paul Dolan) writes:
)>The solution: how about trying it and seeing how much stuff breaks.
)
)Peter has a very good idea here, which I think a couple of others have also
)said one way or another. Let me add a suggestion. Since it looks like 1.4
)is 6 months to a year away, how about making this a widely distributed beta
)release (like to everyone) that does something gross like write "beta
)release" in two inch high letters across the bootup screen, and accompany
)it with a note saying "we think this is going to break a lot of software;
)please try it and report problems that look like <description> to your
)third party software vendor and to <commodore contact>; THIS TEST RELEASE
)IS NOT KNOWN TO BE SAFE TO USE WHILE DOING OPERATIONS ON VALUABLE DATA,
)KEEP EXTRA FREQUENT BACKUPS; don't discard your 1.2 or 1.3 workbench disks
)yet."
Hell, I think we should put that notice on it when we ship final release.
)Just a stray thought.
)Kent, the man from xanth.
jimm, stray system programmer
PS: 6 months to a year? That makes my stomach hurt.
--
Jim Mackraz, I and I Computing
amiga!jimm BIX:jmackraz
Opinions are my own. Comments regarding the Amiga operating system, and
all others, are not to be taken as Commodore official policy.cmcmanis%pepper@Sun.COM (Chuck McManis) (06/17/88)
In article <5580@xanth.cs.odu.edu> kent@xanth.UUCP (Kent Paul Dolan) writes: > ... how about making this a widely distributed beta >release (like to everyone) that does something gross like write "beta >release" in two inch high letters across the bootup screen, and accompany >it with a note saying "we think this is going to break a lot of software; >please try it and report problems that look like <description> to your >third party software vendor and to <commodore contact>; THIS TEST RELEASE >IS NOT KNOWN TO BE SAFE TO USE WHILE DOING OPERATIONS ON VALUABLE DATA, >KEEP EXTRA FREQUENT BACKUPS; don't discard your 1.2 or 1.3 workbench disks >yet." >Kent, the man from xanth. Unfortunately it wouldn't work. It would be almost OK to just release it to all registered developers (Commercial and Certified) because they know what "beta" means. Unfortunately, users have no idea of what the significance of that word is. Just the other day at the local computer store where some user was surreptiously bragging that he already had 1.3, he told me that the internal name for the workbench at Commodore was "Gamma Workbench." And, that they were not releasing it until the new chips were out. He assured me it was already done (wink, wink, nudge, nudge). The other problem with this scheme is that by asking for feedback on what it will break, you imply you will fix it so that it doesn't break those things. And that is probably a false hope. Anyway, all I can say is that I hope the port to the 68020 with MMU has an option in it to protect a task that a developer writes (RunProtected maybe) so that a developer can be *sure* that his/her program is not doing anything that damages the rest of the system. In so doing, the reliability of all Amiga's goes up because the code they run is much more well behaved. And the development cycle speeds up because those same developers don't have to reboot when their code goes awry. --Chuck McManis uucp: {anywhere}!sun!cmcmanis BIX: cmcmanis ARPAnet: cmcmanis@sun.com These opinions are my own and no one elses, but you knew that didn't you.
dillon@CORY.BERKELEY.EDU (Matt Dillon) (06/17/88)
>things. And that is probably a false hope. Anyway, all I can say is that >I hope the port to the 68020 with MMU has an option in it to protect a >task that a developer writes (RunProtected maybe) so that a developer >can be *sure* that his/her program is not doing anything that damages >the rest of the system. In so doing, the reliability of all Amiga's goes >up because the code they run is much more well behaved. And the development >cycle speeds up because those same developers don't have to reboot when >their code goes awry. The problem isn't necessarily that the task is writing over all of memory, but that it might be giving standard library routines bogus arguments. Which makes the MMU almost useless unless major argument checking is put into the library routines, which can be done without using an MMU. ALL the MMU protects us from is a program stomping over memory it isn't supposed to stomp over. I'm not saying an MMU will catch none of the crashes, just that it won't catch all of them (if you can make sense out of that!) -Matt