Hacker News Reader

Comment by PaulHoule | Hacker News Reader

OS 8 and 9 were unreliable at using the internet because unlike Windows or Linux it was not a 32 bit OS with real memory protection. I remember various releases that struggled with the problem, especially they added a whole bunch of locks to protect memory structures which slowed the machine down dramatically.

Classic Mac OS was designed to handle events raining in from the keyboard and mouse and was able to run on the 128k classic Mac but Apple was in serious trouble in the 1990s until MacOS X which was competitive with the competition.

View on HN

>> unlike Windows or Linux it was not a 32 bit OS with real memory protection.

Don't forget that until Windows 2000, neither was Windows

How's that? Earlier versions of Windows NT had memory protection (starting with NT 3.1 in 1993.)

I guess the average consumer was on Windows 95 or 98, though, right?

I ran NT4 on my laptop in the late 90s and remember it was somewhat unusual. A little clunky, my memory is some common things like changing screen resolution required a reboot, but it was more reliable and responsive overall. Most people stuck with the 95/98/Me progression due to cost (2000 wanted more memory than Me and partially for that reason Me was default preload on most PCs).

I ran NT4 on my Thinkpad briefly around 98 or 99... it sucked. Power management didn't work at all. USB didn't work at all. The video driver was awful. The built-in ethernet didn't work; you had to use a PCMCIA card to get networking at all.

Essentially, it turned the laptop into an under-powered desktop with a built-in UPS and slow video.

These things were generally better by the time 2000 came around... but NT4 was never any good on my laptop. (And at the bank where I worked when I graduated in 1999, it was terrible on those as well. The senior managers' laptops were the only systems in the bank allowed to use 98.)

NT 4 was rubbish for laptops. But it was a state-of-the-art OS for desktops in 1996.

Laptops were not so crucial then. Few people used a laptop as their primary computer; they were an adjunct to a _real_ computer, suitable for emergency working on the move, but with poor CPU, slow disks, poor graphics, etc.

The default computer for business until ½ decade or so into the 21st century was a desktop PC, and for them, from 1993 onwards, NT was unrivalled.

And I speak as someone who spent their own money on OS/2 2.0. I've been working in the industry since the 1980s, which means I never spent my own cash on software. (I'm not a gamer.) I've been using FOSS for 25+ years now, but OS/2 2.0 was worth the money.

> NT 4 was rubbish for laptops. But it was a state-of-the-art OS for desktops in 1996.

That matches my recollection. I was working in a shop that used NT 4 on 15000+ desktops. (I do recall SP3 or SP4 causing a ton of drama on those.) We were never able to get it to run acceptably on a laptop.

I never used OS/2 2.0 in anger. I did run Warp 4 on a few systems and recall preferring it over NT 4. It was also only suitable for desktops, though.

Indeed. Back in the days when RAM was hideously expensive, Windows NT needed too much RAM for the home PCs of the era.

Also, Windows 9x had plug and play and the NT family did not.

Most home users didn't switch over to the NT family until Windows XP.

Probably! I was in college at the time. All the fellow nerds either ran Linux or NT.

Windows 95 absolutely had memory protection and preemptive multitasking, at least for “native” Win32 applications.

System 7 had neither.

> Windows 95 absolutely had memory protection and preemptive multitasking, at least for “native” Win32 applications.

I know it had preemptive multitasking, but did it have memory protection? I didn't use Windows in that era, but my recollection is it didn't.

I do remember all the apologetics for cooperative multitasking, because that's what Macs were stuck with so it had to be justified.

Win32 processes were isolated. The problem is that certain kernel objects were limited and could be leaked leading to stability problems. Compatability with DOS required some regions of memory to be unprotected.

> Windows 95 absolutely had memory protection and preemptive multitasking, at least for “native” Win32 applications.

Apparently the first 1MB of RAM was mapped for compatibility with DOS and 16-bit Windows, which meant that the common bug of writing through a null/zero pointer could easily crash the system.

Sort of... but enough bits of the system were still 16 but and it crashed every bit as much as the Mac

Wrong.

Windows 2000 is NT version 5.

In 1993, NT 3.1 was a full 32-bit OS with memory protection.

So was NT 3.5, NT 3.51, and NT 4...

And seven years later the 5th version, Windows 2000.

I mean, I worked for years as a professional web developer on MacOS 8 and 9 machines. The stability wasn't great, but it was fine.

Same history here, but to be honest, a good day with Mac OS 7/8/9 was literally defined by having no forced reboots due to a crash. Especially somewhat heavy multitasking with Photoshop, BBEdit, Netscape and Eudora usually lead to a system crash once every few hours. You can tell the seasoned Mac user by them having seconds enabled on the menu bar clock, so they can see whether the machine is still doing anything :)

With later hardware and software, especially G3/G4 with Mac OS 9.x, the machine crashed less often, but usually still did at least a few times per week.

You could have stopped and started your list with “Netscape” causing crashes. IE for Mac was a “glass of ice water in hell” compared to Netscape.

Yes, for a while IE was the best browser for Mac, not just in terms of stability. But before that time, we had to make do with Netscape, as there wasn't really any worthwhile competition.

For a long time, the Mac version of IE 5 was one of the best browsers on the market. It used a completely different renderer from the Windows version; as a result, its CSS support was quite good. It even supported transparent PNGs, which the Windows version of IE struggled with for years.

> Same history here, but to be honest, a good day with MacOS 7/8/9 was literally defined by having no forced reboots due to a crash.

I recall having unexpected reboots every few days (except on systems with complicated SCSI chains, what a nightmare), but my bar for system extensions was also unreasonably high. I remember feeling a little sick seeing multiple rows of extensions when friends restarted their Macs.

I gave up the "Oscar the Grouch" system extension where said Oscar sung a little song on emptying the Trash only reluctantly, but doing so did make my system more stable.

> Photoshop, BBEdit, Netscape and Eudora

This exact combo, plus MacSSH, Ircle and SoundJam, for years.

Of course, I only mentioned the essentials :) I wasn't much of an IRC person, and I remember using a different SSH client, but don't recall which one. Also, Fetch.

Well now we are enemies! I was Anarchie over Fetch all the way. Really all of the Stairways Software apps were tops.

same, but I pissed off Onno when I pirated Ircle, and ended up going with a different program. ShadowIRC. :) still friends on Facebook with the developer all these years later.

Well we all pirated it, but I think actually I ended up paying for it in the end. I did spend some time on ShadowIRC. I have a vague recollection that that author hung out on Undernet and his website had an animated gif of the moon that I spent WAY too many hours trying to duplicate in PhotoShop. DShadow?

yep, that's the guy. the channel #Macintosh is still there and gets the occasional user logging in from one of the old clients.

Netscape, in general, was incredibly unstable. I remember 3.x crashing every half hour.

I recall OS 8 crashing often and loosing lots of work. Back then I was doing graphic design work and just getting into web programming. Photoshop, Illustrator and QuarkXpress all day with those apps crashing unexpectedly. MS Word for Mac was awful back then too. OS 9 was much better and Windows 2000 was actually really good. It's great that that sort of instability is a thing of the past!

"Save early, save often!"

As another respondent also remembered it, I would probably get less than one crash a day. Windows 95 had it's own (multifarious) problems. I was maintaining servers with years of uptime and definitely looking forward to OSX. :)

> OS 8 and 9 were unreliable at using the internet because unlike Windows or Linux it was not a 32 bit OS with real memory protection.

This isn't my recollection at all, although the lack of protected memory on "Classic Mac OS" means that you would've personally experienced unreliable anything if you were also running poorly-written extensions or other software on your system.

System 7 (1991) was the first 32-bit version of Classic Mac OS, and Internet Explorer for Macintosh was very advanced for its time. Classic Mac OS had a healthy browser ecosystem which also included Cyberdog, iCab, Netscape, Opera, and WannaBe (a text browser).

Seems like they conflated 32-bit with lack of memory protection.

(Tangent: IIRC there were Macs where the CPU had a 32 bit pointer width but you could only use 24 bits of address space. And sometimes software would work on this assumption and break with a full 32 bits. But that's neither here nor there.)

But in addition to lack of memory protection, cooperative multi-tasking was also a big problem. It was a pretty common experience to see something do intense work rendering a progress bar and the rest of the system would slow down. IIRC this made having something like a web browser open sometimes painful.

Woz has an interesting section in his book "iWoz" regarding System 7 stability before/after Internet Explorer was installed. A fresh install of System 7 would operate just fine, but as soon as IE was installed, weird things would happen. It's definitely worth reading if you haven't already.

I was curious and looked up iWoz, and it seems to be a little mixed-up on the dates of things since it talks about using iCab two years before it was available: https://datassette.nyc3.cdn.digitaloceanspaces.com/livros/iw... (Book page 296)

"That first day I used iCab instead of IE, I had no crashes. Not a single one. Hmm." "I could never convince Apple. This was such a big lament for me at the time. I couldn’t convince anyone that it wasn’t the Mac OS that was at fault. Then one day Gil Amelio told me that Apple—in addition to avoiding excess production and inventory and keeping expenses down—was going to buy a new operating system."

The NeXT merger was publicly announced in February 1997, so that private conversation with Amelio about their intent most likely would have been some time in 1996: https://www.tech-insider.org/mac/research/1997/0207.html

iCab, on the other hand, didn't ship until February 1999 as a time-bombed beta: https://web.archive.org/web/20020305110041/http://advergence...

I wonder if he was experiencing instability due to Code Fragment Manager instead of IE itself? Mac IE was even late shipping for 68K Macs due to problems with CFM-68K.

"Finally, Internet Explorer has been hit by the CFM-68K bug. Microsoft was on the cutting edge in adopting the Code Fragment Manager for 68K Macs, and due to the well-known bug in CFM support on 68K Macs, Internet Explorer is currently only available for PowerPC-based Macs. Apple should have this bug fixed soon, though, and Microsoft plans on releasing a 68K version at that time." — https://www.macobserver.com/reviews/ie3.shtml

CFM-68K was infamously unstable until version 4.0 shipped in April 1997, and even the PowerPC-native version saw a bunch of improvements around that time, so this lines up with a 1996/1997 timeframe in iWoz if one ignores the anachronistic iCab story:

"A revised Code Fragment Manager that helps some large, PowerPC-native applications launch faster and enables some applications to launch in low memory situations" — https://www.cnet.com/tech/tech-industry/apple-releases-syste... (September 1996)

"In late November 1996, Apple announced a bug in the CFM-68K Runtime Enabler extension. This bug could cause random crashes and hangs, resulting in application instability and potential loss of data. Because of the potential seriousness of these problems, Apple recommended that customers disable the extension. Also, Mac OS 7.6 would prevent the extension from loading. Mac OS 7.6 does support the the 4.0 version of the extension." — https://macgui.com/kb/article/502 (May 1997)

> Classic Mac OS had a healthy browser ecosystem which also included Cyberdog, iCab, Netscape, Opera, and WannaBe

Ehh...

Cyberdog was never widely used. It depended on the OpenDoc framework, which most users didn't even have installed, and was only available as a beta for a year or so. It was never a serious contender.

iCab was a latecomer -- it was released in 1999, long after most of the mainstream browsers. It was used by a few people, but was never particularly popular either.

I have never even heard of WannaBe. Based on what I see online, it looks pretty obscure.

I absolutely loved Cyberdog. My browser and e-mail client of choice

Having fully indexed instantaneous search for my emails in 1996 was absolutely living in the future.

Especially if you subscribed to a lot of mailing lists and went looking for information on something specific that you knew you had read, but didn't know where.

Yes. It was a sad day when I had to move on from it. Curious as to other applications which were way ahead of the curve (and the hardware) which could be remade today & fly.

To memory what really messed with OS stability when browsing the internet on System 7.5.3 (the first Mac OS I used) up through 9.2.2 were browser plugins.

Earlier on, Java applets were by far the most notorious troublemakers in this realm. I remember as a kid bumping around the internet on a Performa 6400 with Netscape looking for online games, and that machine could never go long without the browser crashing or the dreaded "Sorry, a system error occurred." dialog popping up after browsing a few pages with applets.

Shockwave and Flash were reasonably stable for the most part, until 3D Shockwave games started appearing… those were almost as messy as applets were.

> To memory what really messed with OS stability when browsing the internet on System 7.5.3 (the first Mac OS I used) up through 9.2.2 were browser plugins.

OS Extensions in general were a source of instability. It's like kernel extensions on Macs or kernel drivers on Windows today, there is no safety net and coding errors can take down the entire system.

You really needed to weigh the reputation of a company for stability before adding an extension from them.

> You really needed to weigh the reputation of a company for stability before adding an extension from them.

And the interaction between different extensions. You could be fine with some set of extensions, fine with a different set of extensions, but then have an unusable system if both sets were active.

It didn't have memory protection but it was 32 bit from the start and most apps were “32 bit clean” by version 7.5 or so.

Reply on news.ycombinator.com