The history section is depressing: a whole conga line of promised collaborations and technologies that Apple seems to have promised, and then completely dropped the ball on. (Even apart from the famous crash-and-burns of Copland and OpenDoc, don't forget A/UX 4.0! A/UX and AIX on PowerOpen! Mac OS 7 on DR-DOS! Netware on PPC Macs! Mac OS on the ANS! Okay, wait, MAE on AIX on the ANS! Whoops, no SMP on the ANS! Okay, now we're shipping Mac OS, A/UX, and NT on ANS!)
It puts contemporary Apple's NIH complex in context.
The huge list of projects from that era of Apple really shows how aimless they were.
I remember at the time how more and more irrelevant they seemed compared to the Wintel behemoth, and that really didn’t start to turn around until the G3 iMac, and later the iPod.
>The huge list of projects from that era of Apple really shows how aimless they were.
Most of those projects involved active collaboration with other companies who were similarly aimless, so Apple was by no means alone in their aimlessness.
It's truly wild just how much time, effort, and money the various companies spent smelling each others' farts back in the 90s.
The era was one of fits & starts, mainly because the basic things which could be done were done already, & pretty well — word processors, spreadsheets, databases, page design, software which fit needs, was powerful enough, & wasn't too buggy to use. The long hardware stall of the 80s (think about how the 6502 lasted, basically the same, for well over a decade, & how IBM was still selling an 8086 machine in '87) forced software makers to focus on quality products. What wasn't there yet, even with the improvements through the 90s, was enough grunt in the hardware to do the things that were significantly past those basics, & so there was a lot of "let's try this, let's try that" throwing things at the wall. There were great ideas, & shots at getting them right — contemporary interfaces still look sad beside NeXTSTEP, & the Newton wasn't approaching what it was supposed to be until near when it was axed. Companies don't just want to do bug fixes & incremental improvements; they want their customers to be excited about something, & the 90s churn had a lot to do with keeping people interested & invested.
Pity as SOM was much better alternative than COM, and OS/2 in general.
However that wasn't the only issue that nailed OS/2's destiny, higher hardware resources and lack of proper management direction from IBM side also played a big role.
And clearly a leadership/partnering-driven madness. They had the answer all along: A/UX, a MacOS UX on a Unix base, gave them what they wanted to get to (modern memory management, stability, multi-user security with a MacOS AI), and what they'd end up buying with NeXT.
Instead at the time I was working at an Apple reseller and we were treated to hugely over-optimisic promises for MacOS 8 that included a Microkernel that could run Windows apps alongside MacOS applications.
I believe it is a reference to the infamous Star Trek project. I did not know of the Digital Research connection, but it sounds like it was another one of those collaborations.
The funny thing is, the industry knew that collaboration would be necessary back then, but it took open source models for it to become effective. (I guess direct business to business agreements made competitors to dependent upon each other.)
(author) Yes, exactly. Star Trek got a lot of heat at the time for being a bad idea and at the time it probably would have been a real strategic blunder (Spindler was right to kill it IMHO), but it really worked, and Marklar ended up being analogously the same thing.
The biggest problem with Star Trek was that it didn't run existing Mac software, and you couldn't emulate the 68k ISA reasonably on x86 at the time. The key piece of Mac OS X on Intel a decade later was translation software for PowerPC binaries, so that your existing software still worked.
This is a great and accurate historical history of the Network Server and its software.
I worked at Apple, in the same building with the Apple’s Shiner team and AIX teams, and had a Network Server in my office because I was on the team at Apple porting NetWare to the machine—I still have a bootable CD. The project was cancelled just as it was entering Beta when disks were to get into customers hands.
It looked just like Netware except setup was actually more simple as it auto detected stuff on the network simplifying the process. At the time it was being developed for MIPS, PPC, SPARC, and PA-RISC (I’m thinking there was a fifth platform, or maybe that was X86?)
Netware dropped the other RISC processors and until it was cancelled focused mainly on PPC and X86. Apple did all the PPC drivers, and all the device dependent code as well as a modified version of setup. It was dropped as both companies were going into a death spiral about the time Micheal Spindler left and Gil A started. We were told it was a “mutual” decision by both companies but I don’t have any insight into the reason for cancellation beyond that.
Internally we used IBM’s RS6000 compiler. Novell contracted with Cygnus for a GNU based compiler for their NLM developers that was being developed concurrently with our development. That compiler was for all the PIN based architectures (PA-RISC, SPARC, MIPs, PPC) that PIN was initially intended for.
C code (complied, maybe?, with a C++ compiler for things like variable scoping. I forget.) NetWare’s code almost assuredly was pure C as some of the processors it supported may not have had C++ compilers then? Compilation was done on an IBM RS6000, the same as for the Mac OS at the time, but debugging used a Mac running the MPW debugger (I forget the name) connected to another Mac that ran the code. The Metroworks compiler was just becoming a viable around the time the project was cancelled but all internal development, including MasOS used IBM machines as compile servers.
A tiny portion of the code by necessity was written in assembly (which I happened to work on) for thread switching and the page table.
Only Novel’s most common NLMs (apps) were available and no 3rd party ones were available at the time. The Beta would have gone to some number of NLM developers had it not gone out. Somewhere is an article about developing NLMs, I’ll try to find it and post a link.
Interesting to think about a world where Apple was able to embed AIX with PPCToolbox running on the same machine with classic MacOS essentially as the window server. Assuming they could get around licensing cost issues, this might have been a viable path forward for a modern OS in an alternate future.
There was also Latitude porting toolkit, which was a sort of reverse engineered MacOS toolbox that mapped Macos calls to native Unix calls, including Rhapsody, at the source level. Adobe used it to port Illustrator and Photoshop to Solaris & Irix
I had one of these and ran Yellow Dog Linux on it. Linux supported all of the hardware very well including hotswap SCSI and the 20x4 LCD with a nice multi page mini 'top'.
Physically, it was amazing. No tools required, and the processor board rolled out on rails.
> A/UX, Apple's own Unix with a bolted-on Mac compatibility layer
I used a Mac running A/UX circa 1990-91. It worked rather well for the period. I was really surprised when Apple dropped it because it solved a lot of what I saw as problems in MacOS.
The article also mentions Apple talk. There was some free software called UAB, Unix-to-AppleTalk Bridge around that same timeframe.
It puts contemporary Apple's NIH complex in context.