-
Hello MLAers! We've re-enabled auto-approval for accounts. If you are still waiting on account approval, please check this thread for more information.
OpenStep 68k?
- Thread starter moving2
- Start date
Do you mean something like GnuStep? There never was a version of real "OpenStep" that ran on 68k Macs; there was NeXT/OpenStep for 68k NeXT workstations, and variations of OpenStep ranging from a full OS (IE, "OpenStep for Mach" that ran on a small selection of x86, SPARC, and HP PA-RISC hardware) to some application/API-level versions that ran on top of Solaris and Windows NT. (And of course eventually the whole thing morphed into "Rhapsody", and from there into OS X.)
In principle at least you might be able to get GnuStep running on a 68k Mac using NetBSD as the base OS, but it's gonna be a long slog for not a lot of returns.
In principle at least you might be able to get GnuStep running on a 68k Mac using NetBSD as the base OS, but it's gonna be a long slog for not a lot of returns.
When it came to OS development Apple was a hot mess back in 1993: "Pink/Taligent" was falling apart (and the successor project, Copland, likewise turned into a debacle], and the PowerPC transition was sucking up tonnes of engineering talent just making the old system work seamlessly on the new hardware. (And of course there were those other interesting dead ends like A/UX floating around and muddying things even further) If you'd tossed NeXT into the mix back then it almost certainly would have fallen down a black hole of suck and been lost forever.
One of the major factors holding back Unix-based systems at the time was the extreme licensing costs. In 1992 NeXTStep 486 cost $995, which was a fairly typical price for a PC Unix distribution. (In fact it was somewhat on the cheap side.) NeXT might have been able to sell it for *somewhat* less, but all versions of NeXT/OpenStep before they started calling it "Rhapsody" were based on Pre-BSD 4.4 lite UNIX code and technically were bound to pay royalties to AT&T. Rhapsody was basically the alpha/beta testing period during which all the NeXT-specific code was forklifted over to a new royalty-free base OS distribution which borrowed heavily from NetBSD and FreeBSD. (They retained the Mach microkernel, but there to they had to "sanitize" all the low-level interfaces.) I of course have no idea how much each NeXT station with the OS bundled cost in licenses, maybe it wouldn't have necessarily been "prohibitive", but when you factor in the system requirements for NeXT to run well (8MB was a bare, bare minimum in an age where Apple's highest-volume machines were LCs with 2-4MB of RAM and hard disks far too small) and the presumptive requirement to incorporate backwards compatibility with the legacy OS it's extremely unlikely any NeXT-based OS would have appeared before 1997-98 at the earliest.
One of the most critical innovations during the transformation of OpenStep into OS X was the Carbon APIs, which gave developers for the Classic OS a method for forklifting their applications over to run on the new OS "natively" with relatively minor changes compared to rewriting it entirely for OpenStep. OS X almost certainly would have failed if not for the Carbon strategy; it certainly might have been possible for Apple to do a quick shovelware port of NeXT onto the early Power Macs (it *never* would have run at an acceptable speed on any 68k Macs; it was generally considered sluggish on even the best NeXTstations) in which the Classic OS would have run in a "penalty box" similar to Classic but it probably would have gone over like a walrus pole-vaulter. Say what you like about Windows 95, it actually for the most part ran legacy Windows programs *better* than the previous versions.
One of the major factors holding back Unix-based systems at the time was the extreme licensing costs. In 1992 NeXTStep 486 cost $995, which was a fairly typical price for a PC Unix distribution. (In fact it was somewhat on the cheap side.) NeXT might have been able to sell it for *somewhat* less, but all versions of NeXT/OpenStep before they started calling it "Rhapsody" were based on Pre-BSD 4.4 lite UNIX code and technically were bound to pay royalties to AT&T. Rhapsody was basically the alpha/beta testing period during which all the NeXT-specific code was forklifted over to a new royalty-free base OS distribution which borrowed heavily from NetBSD and FreeBSD. (They retained the Mach microkernel, but there to they had to "sanitize" all the low-level interfaces.) I of course have no idea how much each NeXT station with the OS bundled cost in licenses, maybe it wouldn't have necessarily been "prohibitive", but when you factor in the system requirements for NeXT to run well (8MB was a bare, bare minimum in an age where Apple's highest-volume machines were LCs with 2-4MB of RAM and hard disks far too small) and the presumptive requirement to incorporate backwards compatibility with the legacy OS it's extremely unlikely any NeXT-based OS would have appeared before 1997-98 at the earliest.
One of the most critical innovations during the transformation of OpenStep into OS X was the Carbon APIs, which gave developers for the Classic OS a method for forklifting their applications over to run on the new OS "natively" with relatively minor changes compared to rewriting it entirely for OpenStep. OS X almost certainly would have failed if not for the Carbon strategy; it certainly might have been possible for Apple to do a quick shovelware port of NeXT onto the early Power Macs (it *never* would have run at an acceptable speed on any 68k Macs; it was generally considered sluggish on even the best NeXTstations) in which the Classic OS would have run in a "penalty box" similar to Classic but it probably would have gone over like a walrus pole-vaulter. Say what you like about Windows 95, it actually for the most part ran legacy Windows programs *better* than the previous versions.
Display PostScript was pretty big overhead. Macs were fairly cheap systems design wiae, not really comparable to the pricier Nexts with DSPs and all the other fixins.
The most significant difference between the 68k NeXT hardware and Macintoshes was that the former made extensive use of DMA, with dedicated channels for basically every peripheral and several "memory to memory" channels which are vaguely comparable to the "blitter" hardware in Amigas. 68k Macintoshes on the other hand are comparatively "stupid" and make use of programmed I/O for peripheral communication. (Heck, on early Macs and with regard to devices like floppy disk controllers you're lucky if I/O is even interrupt driven vs. cycle-wasting polling.) On a single threaded number-crunching benchmark a 40mhz IIfx certainly might perform better than a 25mhz NeXT Cube, you're basically just counting cycles, but a port of Next/OpenStep to the IIfx would by necessity have to waste a comparatively high number of cycles "manually" shoveling data around paths that are accellerated in the NeXT hardware so overall system performance would very likely be worse.
Also, we sort of have to do some calibration here of what time period we're talking about. NeXT actually discontinued their 68k hardware in 1993, which was floated as the acquisition date, because it no longer could seriously compete with other Unix workstations based on Sparc/MIPS/PA-RISC/etc CPUs. (Also note this was the year the Pentium became officially available. I recall reading an article in a 1993 magazine comparing various "Unix workstations" and in said article they compared, among other things, a Pentium-based PC running SCO Unix, to a Quadra 800 with A/UX; the authors were enthused about A/UX from a feature standpoint but on their benchmarks the system was by a significant margin the slowest one in the lineup, being as much as a full order of magnitude slower than some of the competitors. A NeXTstation was *not* included as the hardware had already been EoL'ed.) If NeXT had fallen into Apple's lap at this point there would have been pretty much zero justification to waste the effort porting the OS to hardware that was already behind the curve. To *seriously* imagine NeXTStep on Mac hardware you pretty much have to push that date back even further, like 1990 or 1991, but even in that case it seems to me more likely that the result would be a NeXT hardware design relabeled as an Apple product with a MacOS virtual machine (like the one from A/UX) sort of shoehorned into the OS. The chances they'd try backporting it run on "normal" Macs don't seem that high because, again, the performance would really suck.
And of course the other problem this introduces into the timeline is it puts Apple in the position of having to pull off both a processor architecture migration (the writing was truely on the wall for 68k, this is why the other UNIX workstation vendors were all fleeing from it as well) *and* an OS migration at the same time. NeXT itself wasted significant resources on Motorola 88k and PowerPC hardware prototypes before opting to pull out of the hardware business entirely so they were fully aware that 68000 was a dead end. Apple already had its own ideas about creating a powerful RISC-based workstation as part of the Pink/Taligent collaboration during this period, and that ended up going nowhere; all that came out of it was the idea of running the classic OS in emulation on a RISC CPU, which resulted in the fast-but-dumb Power Macintoshes of the mid-1990's. It's really hard to imagine how a NeXT acquisition during this period could have resulted in anything but more confusion and dead-ends.
(I'm sure the answer to that is "But, Steve Jobs would provide the vision to make all work!", but, well, I don't think at this point in Steve's life he would have been particularly interested in the job. Both Apple AND NeXT pretty much had to *fail* in order to set the stage for what happened upon his return in 1997, back in 1990 the "synergy" just wouldn't have been there.)
Also, we sort of have to do some calibration here of what time period we're talking about. NeXT actually discontinued their 68k hardware in 1993, which was floated as the acquisition date, because it no longer could seriously compete with other Unix workstations based on Sparc/MIPS/PA-RISC/etc CPUs. (Also note this was the year the Pentium became officially available. I recall reading an article in a 1993 magazine comparing various "Unix workstations" and in said article they compared, among other things, a Pentium-based PC running SCO Unix, to a Quadra 800 with A/UX; the authors were enthused about A/UX from a feature standpoint but on their benchmarks the system was by a significant margin the slowest one in the lineup, being as much as a full order of magnitude slower than some of the competitors. A NeXTstation was *not* included as the hardware had already been EoL'ed.) If NeXT had fallen into Apple's lap at this point there would have been pretty much zero justification to waste the effort porting the OS to hardware that was already behind the curve. To *seriously* imagine NeXTStep on Mac hardware you pretty much have to push that date back even further, like 1990 or 1991, but even in that case it seems to me more likely that the result would be a NeXT hardware design relabeled as an Apple product with a MacOS virtual machine (like the one from A/UX) sort of shoehorned into the OS. The chances they'd try backporting it run on "normal" Macs don't seem that high because, again, the performance would really suck.
And of course the other problem this introduces into the timeline is it puts Apple in the position of having to pull off both a processor architecture migration (the writing was truely on the wall for 68k, this is why the other UNIX workstation vendors were all fleeing from it as well) *and* an OS migration at the same time. NeXT itself wasted significant resources on Motorola 88k and PowerPC hardware prototypes before opting to pull out of the hardware business entirely so they were fully aware that 68000 was a dead end. Apple already had its own ideas about creating a powerful RISC-based workstation as part of the Pink/Taligent collaboration during this period, and that ended up going nowhere; all that came out of it was the idea of running the classic OS in emulation on a RISC CPU, which resulted in the fast-but-dumb Power Macintoshes of the mid-1990's. It's really hard to imagine how a NeXT acquisition during this period could have resulted in anything but more confusion and dead-ends.
(I'm sure the answer to that is "But, Steve Jobs would provide the vision to make all work!", but, well, I don't think at this point in Steve's life he would have been particularly interested in the job. Both Apple AND NeXT pretty much had to *fail* in order to set the stage for what happened upon his return in 1997, back in 1990 the "synergy" just wouldn't have been there.)
Last edited by a moderator:
To be honest, the IIfx did have dedicated IOPs on some hardware including the floppy, but only the IIfx obviously.
Last edited by a moderator:
Well, that's the point, isn't it? The best thing Apple had on the table in 1990 is arguably substantially inferior to the 1988 incarnation of the NeXT hardware. (Also of note is the fact that the 68040 versions of the NeXT computers hit the market almost a year earlier than the Quadra 700/900.) A more "typical" Mac of the era was, again, something like an LC or IIci. (And your average hard drive was something like 40-120MB; the original NeXTStep 1.0-targeted hardware had a 256MB optical drive, and the system requirements for 1992's NeXTstep 486 3.3 recommended at least 330MB for a "stand alone" system.) The cost to upgrade an existing Mac to something like 16MB of RAM and a sufficiently large hard drive would have been somewhere in the $1,500 range in 1992 (the RAM, oddly enough, would be even more expensive in 1993 because of an industrywide shortage), and, again, they would have run the OS pretty badly.
(NeXT/OpenStep actually had a reputation for being somewhat sluggish even on the genuine NeXT hardware; part of the blame goes to the really ill-conceived idea of using the optical drive for the OS in the base configuration, but even with a hard disk it was simply a little bit ahead of its time in terms of system requirements.)
So, yeah, I just don't see it working. OS/2 2.x debuted in the 1992-ish timeframe and its "8MB to run well" system requirements were considered formidable for an OS aimed at the mass market. (Windows 3.0 would run mostly-okay with as little as one or two MB.) Mass market computers simply weren't ready for UNIX-class operating systems until around 1996 or so. (That's about the point that 16MB became "standard"; don't forget, the original iMac, circa late 1998, shipped with only 32MB.) Combine that with the fact that Apple had to suffer through most of 1993 and into 1994 with no systems that could really be legitimately called "workstation class" in terms of CPU power (IE, that uncomfortable gap between the March 1993 introduction of the Pentium CPU and the March 1994 introduction of the first three Power Macs) any attempt to wedge NeXTstep into the product line in 1993 would have been an unmitigated disaster.
(NeXT/OpenStep actually had a reputation for being somewhat sluggish even on the genuine NeXT hardware; part of the blame goes to the really ill-conceived idea of using the optical drive for the OS in the base configuration, but even with a hard disk it was simply a little bit ahead of its time in terms of system requirements.)
So, yeah, I just don't see it working. OS/2 2.x debuted in the 1992-ish timeframe and its "8MB to run well" system requirements were considered formidable for an OS aimed at the mass market. (Windows 3.0 would run mostly-okay with as little as one or two MB.) Mass market computers simply weren't ready for UNIX-class operating systems until around 1996 or so. (That's about the point that 16MB became "standard"; don't forget, the original iMac, circa late 1998, shipped with only 32MB.) Combine that with the fact that Apple had to suffer through most of 1993 and into 1994 with no systems that could really be legitimately called "workstation class" in terms of CPU power (IE, that uncomfortable gap between the March 1993 introduction of the Pentium CPU and the March 1994 introduction of the first three Power Macs) any attempt to wedge NeXTstep into the product line in 1993 would have been an unmitigated disaster.
Similar threads
