liam_on_linux: (Default)
https://emutos.sourceforge.io/

I have an ST, and an Amiga, but I didn't use either back in the day. But I think this is amazing work and really impressive.

So I stuck in on HN and some pillock went "yah boo TOS sucks Amiga is better" like it was 1986. I paraphrase. I am unimpressed.

In fact, while I don't want to be mean, you're missing two or possibly three different points... among which are the reasons I posted this link.

[1] It's not that TOS is less advanced than AmigaOS. Yes it is, and anyone who knows them realises that, but that's not the issue here. The issue is that this FOSS project has brought these two platforms together after about 35 years, and that's both really technologically impressive and also just plain fun.

[2] It means in principle that Amiga owners can run Atari apps, and the ST had some impressive apps.

[3] AROS is great but it's an x86 OS. It doesn't readily run on classic Amigas, or even especially well on the handful of later PowerPC Amigas, AFAIK. It also doesn't run natively on modern RISC hardware, like say the Raspberry Pi.

[4] But because it doesn't, that's prompted the creation of another really cool FOSS project, Emu68 -- a native 68K emulation environment for Arm, something comparable to Apple's nanokernel for running Classic MacOS on PowerMacs.

https://github.com/michalsc/Emu68

[5] Creating an OS that's as good or even better than the original while running on original hardware is impressive. Improved localisation opens it up to more people. That's good. It enables reviving vintage kit more easily, and expanding it. That's great.

You were so busy mocking something that you didn't stop to consider all the good sides.

[6] We know TOS was limited. We all know that. OTOH its simplicity enabled this. Its simplicity also was part of why the ST survived as a musicians' tool of choice for decades after it went out of production: super low latencies for music, and so on.

But others knew that TOS was limited, which drove a 3rd party OS market, with products such as MagiC:

https://en.wikipedia.org/wiki/MagiC

And MagiC is now FOSS:

https://gitlab.com/AndreasK/Atari-Mac-MagiC-Sources

Which is good, but OTOH, it's not attracted much interest or development, AFAICS...

Whereas EmuTOS is now on v 1.21 and is seeing new releases several times a year. This is great, and is one reason I posted it.

[7] The limitations of TOS are also what prompted the development of MINT, and that's FOSS too, and it's quite mature:

https://github.com/totalspectrum/atari-mint

And it has distros, such as AFROS:

https://aranym.github.io/afros.html

Which you can run on x86 kit:

https://aranym.github.io/

All of which is amazing work.

So, yes, while you just wanted to do some advocacy, you missed a huge amount of great work by a committed community.

Not cool, dude.

Leave the Amiga-v-ST hate in the 1980s where it belonged. It wasn't very welcome then. They're both great computers. But hey, then the fans were children, so they can be excused.

In 2022, they can't.
liam_on_linux: (Default)
Something interesting that has come out of Caldera's release of the original DR GEM code as FOSS 20 years ago, and I totally missed it...



This is a great ~40min intro to EmuTOS.

Nowadays there are two different all-FOSS OSes for STs, compatibles & ST emulators.

I knew about AFROS and have played with it -- it's a compilation of various ST GEM enhancements and replacement modules and so on, mostly based on the FreeMINT multitasking OS, to create a complete multitasking GEM OS for advanced STs.

It mainly targets the ARANYM emulator.

The one bit that wasn't free was basically the ST ROM – TOS itself. TOS shared ancestry with both DR's CP/M-68K and what later became DR-DOS. A very rough description is a DOS-like kernel and drivers for the ST hardware, with floppy drive support, just enough to launch the GEM desktop. No command line.

The AFROS project wrote their own ROM, and back when I was actively looking at ARANYM, they described it as something like "just enough ROM to boot our OS, and not very compatible with actual ST software".

Well what I didn't know until this evening is that the EmuTOS project has taken on a life of its own and they released v1.0 about 6 months ago. It's a complete single-tasking GEM replacement for STs: in other words, a whole replacement ROM. It replaces the BIOS and OS kernel and all of the GEM stack, and that part is based on Caldera's GEM code.

They have something that is built in GCC, can just about fit into the smallest ST ROM chip (192kB) and is broadly compatible with Atari TOS 3. For later models it can go into a bigger ROM chip which gives you a command-line and even multi-language support.

Or you can boot it from floppy, or you can load it as an app from real Atari TOS if you have enough memory. You can even boot it on Amigas, with some restrictions currently.

I'm really impressed. I found this very interesting viewing.

Source etc: is on GitHub. There's a slightly dated Wikipedia article too.

There are or were other ST OSes around. A popular one was called MagiC, and at least part of this has been made FOSS recently. It came with emulators to allow it to run on macOS and Windows. Snag: it's largely in assembler, apparently.

But EmuTOS is slightly different from things like AFROS, FreeMINT or MagiC, inasmuch as it's able to run on original unmodified STs (and the Amiga!) and can be freely distributed with emulators.

A company called Atari still exists and still holds the old copyrights, so the original Atari ROMs are not strictly distributable.

Incidentally, I found this via the m68k.info page, which hosted another presentation this weekend, on the Sinclair QL OS descendants Minerva and SMSQ/E.



Not really any relevance to GEM etc. but may be of interest to folk – it was to me.

I found that because I was asking if there were any 16-bit homebrew computers these days, and was told about the amazing Kiwi 68K.
liam_on_linux: (Default)
The evolution of DOS is interesting, and few remember the bigger picture now.

MS did a great deal when supplying DOS to IBM; MS retained the rights to sell it itself to other manufacturers.

So in the early days, there were other MS-DOS machines that weren't IBM compatible, such as the Apricot, Victor and Sirius.

But soon it became apparent that IBM compatibility was key. Compaq reverse-engineered the IBM BIOS and built the first clones, and the PC industry started from there.

PC DOS only came with IBM kit. MS-DOS came with everything else, but only with the computer. You couldn't buy it directly.

Excluding bugfixes, it went like this:

DOS 1: floppy-only machines.
DOS 2: added hard disk support (a single one) and subdirectories.
DOS 3: added support for 2 hard disks and networking. Then in a point release, support for 2 partitions per disk. Then in another point release, multiple "logical drives" in a single extended partition, so you could use all the space on a big drive... but still a max of 32 MB.

Other companies started tweaking their version of MS-DOS 3.3 to allow bigger than 32MB drives. The method used in Compaq DOS 3.31 is the one IBM and MS picked and it was used in DOS 4.

MS had a project to do a multitasking DOS 4 so didn't work on DOS 3.3 for ages. IBM did its own thing, and added big disk support, code page switching for international character sets, and a slightly clunky graphical launcher called DOSShell.

MS reluctantly released this as MS-DOS 4. It's the first release that required a bugfix fairly quickly. The multitasking version got abandoned: big disk support was needed more urgently. But DOS 4 had other gotchas -- such as using a lot more RAM so some apps couldn't run. (Everything in DOS had to fit into the first 640 kB).

DR noticed this. Its CP/M-86 was late, expensive and so lost out to MS, even thought it was the inspiration for SCP’s QDOS, the basis of DOS 1.0. DR had its own line of multitasking CP/M derivatives, for minicomputer like x86 machines with terminals: Concurrent CP/M, and later with DOS app compatibility, Concurrent DOS. It also had its own standalone single-user DOS, DOS Plus, which could run 3 background tasks on a single PC (if they all fitted into what was left from 640 kB after the OS loaded!)

So DR reworked DOS Plus, removed anything that broke compatibility, like the multitasking and CP/M app support, updated its MS-DOS compatibility with code from Concurrent DOS, and released it as DR-DOS. It bumped the version number from the last small, memory-efficient MS-DOS, MS-DOS 3.3, but included compatible large-disk support. So… DR-DOS 3.41.

It only offered it through OEMs at first. You couldn’t buy it at retail. But it proved moderately popular, a sort of cult hit. People heard about it. (This is all in the 1980s so pre-WWW.) People asked to buy it as an upgrade.


So DR had a great idea. There were already 3rd party memory managers for DOS on 386 computers, which let you map RAM into bits of the space between 640 kB and 1024 kB. You couldn’t run bigger apps using this space because it wasn’t contiguous with base memory, but you could load bits of DOS into them: keyboard drivers, CD drivers, mouse drivers, disk caches. Now, instead of having only 500-550 kB of 640 free for your apps after loading all your drivers, you got more room: up to 580-590 kB.

PC/MS-DOS 4 made this even more necessary as it used more memory than DOS 3.3.

DR wrote their own and bundled it into DR-DOS, and leapfrogged MS-DOS 4 by calling it DR-DOS 5. You could even move DOS itself out of the base memory, and have 620-630 kB free, without 3rd party tools. It was amazing. It also added a full-screen text editor, which incredibly MS-DOS still didn’t have.

And in a masterstroke, they made it available at retail. You could buy it in a shop and upgrade your PC or MS-DOS computer.

It sold extremely well and that made MS angry. It had never realised there was a potential retail market in after-market DOS upgrades or additional DOS features; it had been distracted by the success of Windows 3.

So MS copied the features of DR-DOS 5 and, playing catchup, made MS-DOS 5. All the features of MS-DOS 4, more free memory than ever with a memory manager, a full-screen editor (actually part of QBASIC, which was the GW-BASIC interpreter with the IDE from the QuickBASIC compiler.)

And sold it as a retail upgrade.

It did way better than DR-DOS 5 because it had Microsoft’s marketing muscle.

Novell bought DR around this time, intending to go against MS with a multi-pronged strategy: a better DOS, some best-of-breed apps - it also bought WordPerfect, now failing against Windows apps, notably Word of Windows and a Windows port of the Mac’s Excel spreadsheet. To rival Excel it bought Quattro Pro from Borland, a graphical spreadsheet for DOS.

Against Windows itself, Novell planned a Linux-based desktop, codenamed “Corsair”, which eventually became Caldera OpenLinux.

Novell bundled SuperStor disk compression, and re-implemented DOS Plus’ multitasking with TASKMAX.

Result, DR-DOS 6, AKA Novell DOS 6.

Microsoft responded with MS-DOS 6, still playing catchup. It added built-in antivirus and built-in backup, licensed in from other companies who never made the promised monies from selling enhanced versions. It also added disk compression. MS looked at licensing in disk compression from the #1 3rd party vendor, STAC, authors of Stacker. It got to see the code. In the end it didn’t go with Stacker but licensed Vertisoft DoubleDisk instead — presumably because it was cheaper. But it used some Stacker code in DoubleSpace.

STAC sued, won, and spend the money on moving out of the drive-compression market, knowing that drive sizes would grow and make its product irrelevant. It bought the ReachOut remote-control tool, and a server backup tool, and tried to rebrand as a server maintenance tools vendor, foreseeing the rise of internet-based remote admin — but too soon.

The result was MS-DOS 6.1, with no disk compression, while MS rewrote it to remove the stolen code.

Then MS-DOS 6.2, with DriveSpace instead of DoubleSpace, and the SCANDISK improved disk-repair tool.

Then MS-DOS 6.21 and 6.22, bug fixes.

Needless to say, Vertisoft made no money from add-on DriveSpace tools, and Central Point made no money from updates to DOS Antivirus or the bundled PC Backup. Both went under.

Novell responded with DR-DOS 7, with bundled peer-to-peer networking. MS didn’t bother as Windows for Workgroups already included that.

Then MS moved the goalposts with Windows 95, which actually bundled MS-DOS into Windows.

Novell did get Win95 running on top of DR-DOS, but there was no point and it wisely decided not to sell it. Once you had Win95, what DOS did underneath became rather irrelevant, memory management and all.

Novell gave up on the DOS line.

However, the Linux it sponsored did quite well. Caldera was the first desktop Linux I used as my main OS for a while. It had a great setup tool, LISA. It had the first graphical installer. It was the first distro to bundle the new KDE graphical desktop.

It was streets ahead of Red Hat or Debian at the time, let alone Slackware.

So Novell bought the Unix business off AT&T, and SCO, the leading PC UNIX vendor, and tried to get Caldera to integrate these 3 disparate products into a whole and a market.

It didn’t work but that’s a whole other story. What’s relevant to DOS is that Caldera spun off its DOS division as Lineo (who offered me a job once, as a leading DOS expert! But I didn’t want to move to Utah, partly because I like beer, partly because I’m atheist and thought it wouldn’t be too comfortable to live in the Mormon state.)

Lineo tried to make a business out of DR-DOS as a thin client OS. It didn’t work. But Lineo inherited what was left of Digital Research. The Concurrent DOS business had been sold off to 2 of its leading resellers, and that’s just barely still around, amazingly. The realtime OS FlexOS and multitasking X/GEM desktop had been sold off and was sold by IBM until recently, and now by Toshiba.

But the other DR properties — CP/M and the GEM desktop for DOS — Lineo made open source, and both are still around today.

Meanwhile, MS lost interest in DOS as it pursued Windows 95 OSR2, Windows 98 and Windows ME. Indeed the embedded DOS in NT has never moved beyond version 5.5. But IBM co-owns DOS, and it did not lose interest. It continued to develop it for years, including the new features from the embedded MS-DOS within Win9x. The result was IBM PC DOS 7, then PC DOS 2000 (briefly bundled with VirtualPC!) and finally IBM PC DOS 7.1. IBM eschewed MS's editor and BASIC, replacing them with a version of its own OS/2 and mainframe editor E, and replacing QBASIC with REXX. It's an interesting OS.

That is the last ever member of the mighty DOS dynasty. I've blogged about it before. It was never released on its own, but IBM's ServerGuide Scripting Toolkit is a free download and includes the kernel and utilities of PC DOS 7.1. You can combine this with the rest of PC DOS 2000 -- reminder, it was in VirtualPC, and VirtualPC was a free download, too -- and build your own complete working copy. I have it booting "on the metal" on a Thinkpad X200 and it's a pleasure to use -- and very, very fast. Free DOS apps such as Microsoft Word 5.5, the AsEasyAs spreadsheet, the WordPerfect Editor and so on all run fine and amazingly fast.
liam_on_linux: (Default)
More retrocomputing meanderings -- whatever became of the ST, Amiga and Acorn operating systems?

The Atari ST's GEM desktop also ran on MS-DOS, DR's own DOS+ (a forerunner of the later DR-DOS) and today is included with FreeDOS. In fact the first time I installed FreeDOS I was *very* surprised to find my name in the credits. I debugged some batch files used in installing the GEM component.

The ST's GEM was the same environment. ST GEM was derived from GEM 1; PC GEM from GEM 2, crippled after an Apple lawsuit. Then they diverged. FreeGEM attempted to merge them again.

But the ST's branch prospered, before the rise of the PC killed off all the alternative platforms. Actual STs can be quite cheap now, or you can even buy a modern clone:

http://harbaum.org/till/mist/index.shtml

If you don't want to lash out but have a PC, the Aranym environment gives you something of the feel of the later versions. It's not exactly an emulator, more a sort of compatibility environment that enhances the "emulated" machine as much as it can using modern PC hardware.

http://aranym.org/

And the ST GEM OS was so modular, different 3rd parties cloned every components, separately. Some commercially, some as FOSS. The Aranym team basically put together a sort of "distribution" of as many FOSS components as they could, to assemble a nearly-complete OS, then wrote the few remaining bits to glue it together into a functional whole.

So, finally, after the death of the ST and its clones, there was an all-FOSS OS for it. It's pretty good, too. It's called AFROS, Atari Free OS, and it's included as part of Aranym.

I longed to see a merger of FreeGEM and Aranym, but it was never to be.

The history of GEM and TOS is complex.

Official Atari TOS+GEM evolved into TOS 4, which included the FOSS Mint multitasking later, which isn't much like the original ROM version of the first STs.

The underlying TOS OS is not quite like anything else.

AIUI, CP/M-68K was a real, if rarely-seen, OS.

However, it proved inadequate to support GEM, so it was discarded. A new kernel was written using some of the tech from what was later to become DR-DOS on the PC -- something less like CP/M and more like MS-DOS: directories, separated with backslashes; FAT format disks; multiple executable types, 8.3 filenames, all that stuff.

None of the command-line elements of CP/M or any DR DOS-like OS were retained -- the kernel booted the GUI directly and there was no command line, like on the Mac.

This is called GEMDOS and AIUI it inherits from both the CP/M-68K heritage and from DR's x86 DOS-compatible OSes.

The PC version of GEM also ran on Acorn's BBC Master 512 which had an Intel 80186 coprocessor. It was a very clever machine, in a limited way.

Acorn's series of machines are not well-known in the US, AFAICT, and that's a shame. They were technically interesting, more so IMHO than the Apple II and III, TRS-80 series etc.

The original Acorns were 6502-based, but with good graphics and sound, a plethora of ports, a clear separation between OS, BASIC and add-on ROMs such as the various DOSes, etc. The BASIC was, I'd argue strongly, *the* best 8-bit BASIC ever: named procedures, local variables, recursion, inline assembler, etc. Also the fastest BASIC interpreter ever, and quicker than some compiled BASICs.

Acorn built for quality, not price; the machines were aimed at the educational market, which wasn't so price-sensitive, a model that NeXT emulated. Home users were welcome to buy them & there was one (unsuccessful) home model, but they were unashamedly expensive and thus uncompromised.

The only conceptual compromise in the original BBC Micro was that there was provision for ROM bank switching, but not RAM. The 64kB memory map was 50:50 split ROM and RAM. You could switch ROMs, or put RAM in their place, but not have more than 64kB. This meant that the high-end machine had only 32kB RAM, and high-res graphics modes could take 21kB or so, leaving little space for code -- unless it was in ROM, of course.

The later BBC+ and BBC Master series fixed that. They also allowed ROM cartridges, rather than bare chips inserted in sockets on the main board, and a numeric keypad.

Acorn looked at the 16-bit machines in the mid-80s, mostly powered by Motorola 68000s of course, and decided they weren't good enough and that the tiny UK company could do better. So it did.

But in the meantime, it kept the 6502-based, resolutely-8-bit BBC Micro line alive with updates and new models, including ROM-based terminals and machines with a range of built-in coprocessors: faster 6502-family chips for power users, Z80s for CP/M, Intel's 80186 for kinda-sorta PC compatibility, the NatSemi 32016 with PANOS for ill-defined scientific computing, and finally, an ARM copro before the new ARM-based machines were ready.

Acorn designed the ARM RISC chip in-house, then launched its own range of ARM-powered machines, with an OS based on the 6502 range's. Although limited, this OS is still around today and can be run natively on a Raspberry Pi:

https://www.riscosopen.org/content/

It's very idiosyncratic -- both the filesystem, the command line and the default editor are totally unlike anything else. The file-listing command is CAT, the directory separator is a full stop (i.e. a period), while the root directory is called $. The editor is a very odd dual-cursor thing. It's fascinating, totally unrelated to the entire DEC/MS-DOS family and to the entire Unix family. There is literally and exactly nothing else even slightly like it.

It was the first GUI OS to implement features that are now universal across GUIs: anti-aliased font rendering, full-window dragging and resizing (as opposed to an outline), and significantly, the first graphical desktop to implement a taskbar, before NeXTstep and long before Windows 95.

It supports USB, can access the Internet and WWW. There are free clients for chat, email, FTP, the WWW etc. and a modest range of free productivity tools, although most things are commercial.

But there's no proper inter-process memory protection, GUI multitasking is cooperative, and consequently it's not amazingly stable in use. It does support pre-emptive multitasking, but via the text editor, bizarrely enough, and only of text-mode apps. There was also a pre-emptive multitasking version of the desktop, but it wasn't very compatible, didn't catch on and is not included in current versions.

But saying all that, it's very interesting, influential, shared-source, entirely usable today, and it runs superbly on the £25 Raspberry Pi, so there is little excuse not to try it. There's also a FOSS emulator which can run the modern freeware version:

http://www.marutan.net/rpcemu/

For users of the old hardware, there's a much more polished commercial emulator for Windows and Mac which has its own, proprietary fork of the OS:

http://www.virtualacorn.co.uk/index2.htm

There's an interesting parallel with the Amiga. Both Acorn and Commodore had ambitious plans for a modern multitasking OS which they both referred to as Unix-like. In both cases, the project didn't deliver and the ground-breaking, industry-redefiningly capable hardware was instead shipped with much less ambitious OSes, both of which nonetheless were widely-loved and both of which still survive in the form of multiple, actively-maintained forks, today, 30 years later -- even though Unix in fact caught up and long surpassed these 1980s oddballs.

AmigaOS, based in part on the academic research OS Tripos, has 3 modern forks: the FOSS AROS, on x86, and the proprietary MorphOS and AmigaOS 4 on PowerPC.

Acorn RISC OS, based in part on Acorn MOS for the 8-bit BBC Micro, has 2 contemporary forks: RISC OS 5, owned by Castle Technology but developed by RISC OS Open, shared source rather than FOSS, running on Raspberry Pi, BeagleBoard and some other ARM boards, plus some old hardware and RPC Emu; and RISC OS 4, now owned by the company behind VirtualAcorn, run by an ARM engineer who apparently made good money selling software ARM emulators for x86 to ARM holdings.

Commodore and the Amiga are both long dead and gone, but the name periodically changes hands and reappears on various bits of modern hardware.

Acorn is also long dead, but its scion ARM Holdings designs the world's most popular series of CPUs, totally dominates the handheld sector, and outsells Intel, AMD & all other x86 vendors put together something like tenfold.

Funny how things turn out.

May 2025

S M T W T F S
    12 3
45678910
11121314151617
1819 2021222324
25262728293031

Syndicate

RSS Atom

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Jun. 13th, 2025 06:41 am
Powered by Dreamwidth Studios