[Spheric Harlot]

Almost all my crashes were due to Internet Explorer eating into neighboring apps' memory space.

When I made sure not to open anything after opening IE, I'd be alright.

[P]

Originally Posted by msuper69 

No. The crashes were caused by lack of protected memory. Errors in the extensions or an application could easily overwrite a critical area of the OS or another application causing the OS to go into an infinite loop.

This is a bit like saying that a car accident was caused by the driver not wearing his seatbelt. The crash was caused by something else, but the seatbelt would have limited the consequences fo the crash, the same way that protected memory would have limited the consequences of a bad memory access to an app crashing rather than the OS panicing.

Also: System 7 on 68k had protections built-in that would catch most random memory faults without bringing the OS down: It would catch a memory access that was out of bounds, an unaligned memory access, or attempting to execute an illegal instruction. These are errors type 1, 2 and 3, respectively, and statistically they would catch almost all bad memory accesses (they would not catch an app simply overflowing its assigned RAM - there was a feature for that called guard pages, but that was never implemented in a shippping version of the OS). On PPC, this feature did not work until 7.6.1, and all those errors became an unrecoverable type 11 error instead. Stability improved significantly when that feature was ported.

Originally Posted by Spheric Harlot

Almost all my crashes were due to Internet Explorer eating into neighboring apps' memory space.

When I made sure not to open anything after opening IE, I'd be alright.

Interesting. I always thought IE was the more stable of the browsers, as Netscape would die inside minutes if the MMM was on. Fun fact for the historically minded: I know for a fact that Symantec, when looking for disk faults that Norton couldn't repair, received some drives back that had HTML code written all over the disk directory. One browser - not known which, or at least I don't know - sometimes wrote its cache over the disk directory. That's a nice bug.

[Spheric Harlot]

Heh.

It was definitely considered standard practice to make sure nothing else was opened after IE, until IE was quit.

[Simon]

Of course hat shouldn't distract from the fact that Classic Mac OS crashed frequently already long before Microsoft launched IE for the Mac.

[Spheric Harlot]

Actually, my system was fairly stable apart from that.

Required a lot of futzing with Conflict Catcher, though.

[Simon]

Well, Classic Mac OS didn't need MS to crash, that's for sure.

Stable has also changed meaning quite a bit since then. Today stable is no less than total rock solid behavior even when your uptime is months. Back in the day stable meant not losing a day's work when some app crash took down the entire system five minutes before you finished the project and shut the computer off for the night.

[P]

We're pickier now. Back then, you might get angry if you lost a complete document, and you'd just get told that you should have saved. Now, people freak out if they lose their open tabs in the browser.

Crashing due to INITs is also more behaviour than OS design. INITs correspond to kernel extensions, and a kext can certainly crash the OS today. The difference was that everyone had several more or less debugged INITs installed. That combined with the horribly buggy System 7.5 meant that the system became unstable, and none of those bugs would have been helped by protected memory (they might have been easier to analyze, however).