Posts by d.fawcus

How Sinclair's QL computer outshined Apple's Macintosh against all odds

Re: Multitasking vs dates

For Xenix, while it needed a hard disk, it didn't need a 286. See page 8:

https://bitsavers.org/pdf/sco/pc_xenix/XENIX_3.0_for_the_IBM_PC_Release_1.1_Release_Notes_Jul84.pdf

Even before that, there was Veniix/86 demo in May '83, also running on an 8086. Again I believe it needed a hard disk.

(Note that the IBM PC-AT was not released until August '84)

As to Concurrent CP/M (and hence Concurrent DOS) - they did not need hard disks to perform their multitasking. That multitasking was limited, but OEM tunable, plus it did support more than just full screen windows. It supported multiple overlapping resizable text windows.

https://bitsavers.org/pdf/digitalResearch/concurrent_cpm/4007-1010_Concurrent_CPM_with_Windows_Technical_Note_for_IBM_PC_Feb84.pdf

(The rest of your comments stand)

Elsewhere there is mention of a multitasking version of C/PM-68k and of GEMDOS.

Well, there sort of was a multitasking C/PM-68k, but only as a side effect of the CDOS-68k effort. Hence too late for the QL. Plus too big. The system image (CDOS.SYS) is 187764 bytes resident text+data+bss. The text could have been ROMed, but that alone is 161622 bytes.

In addition to its native API, whereas CDOS-286 had an DOS front end, CDOS-68k also came with a C/PM-68k front end. The compiler supplied with its development kit was the C/PM-68k version of the Alycon compiler. Since CDOS-68k used a FAT filesystem, so does the C/PM-68k front end.

As to how well it works, I don't know as I've not yet got it running. Lacking a VME/10 system, and not wanting to write an emulator for one, I'll have to port it to a virtual 68k environment.

As to if a concurrent C/PM-68k could have been produced sooner; given a customer demand, probably - and with lower resource requirements.

Concurrent C/PM-86 was an outgrowth of M/PM-86. Now M/PM existed for 8080/Z80 and 8086, and the v1 spec is quite small. So the core modules could probably have been ported to the 68k (since they're so simple), and the early C/PM-68k source added on top. I believe the first version of C/PM-68k was in assembly (or maybe Pascal), later versions were in C.

As to GEMDOS, yes it had a CLI; it can be seen in its porting kit. I don't know if the ST was supplied with one, but I doubt it - given that a bloke I knew in Uni wrote his own one for it.

Multitasking vs dates

"And remember, the AT also didn't have multitasking. That only came much later, with SCO Xenix, because Intel removed a feature from the final version of the 80286 CPU which DR Concurrent DOS 286 needed to multitask DOS apps."

I'm not sure what date you have for that version of Xenix [1], but CDOS-286 had protected mode pre-emptive multitasking of its own native apps at release - whenever that was [2]. It was inherent in how the system worked. The 68000 port of it (ver 1.2) also had such support, and it was available at latest by August '86 [3]. Someone could try porting that to the "2 bit" QL, or even an Atari ST. The porting kit is available.

Anyway, since you brought Xenix in as a valid comparison for multitasking, you should also include M/PM, Concurrent C/PM, and Concurrent DOS. The former would be '81 (on an 8080/Z80/8086), '82 for CCPM, but more practically for DOS apps only from '84.

(All of these OS dates would have been earlier for various OEM evaluations)

As to a 68008 based business machine at the time. Well it could have worked - if it had come out first, someone recently retro-fitted a 68000 to an 8088 based IBM PC, and found it gave comparable performance. Just that the QL wasn't really up to snuff, as it'd have to have been one with 16 bit memory, to start to compete. Since 8086 based DOS machines were already available, and a machine with a different CPU would have had to be faster.

[1] WinworldPC,and Bitsavers both suggest PC Xenix was available in 1984, possibly July.

[2] Ver 1.3 is dated Dec '86 (Bitsavers). It apparently worked (poorly) for DOS apps with the E-1 Step of the 286 by Aug '85, and better with the E-2 Step. That chip was supposed to be available from "first quarter 1986".

[3] I've seen references to it first being available in '85, possibly June. However not before Jan '84 when the work was announced. A Usenet post from Jun '86 suggests it had been operation for "about a year and a half", so maybe from Jan '85. Which then also suggests that CDOS-286 v1.2 was around (for OEMs) at that same time (sans DOS apps), as it was renamed FlexOS with the 1.3 version.