MS-DOS 4.0 (multitasking)

[dasyar]

I was doing some searching on the internet and came across some items that mentions that there was/is a European MS-DOS 4.0. Since this forum has a lot of European members, has anybody really used this version of DOS?

I also noticed that the software is available, but I do not have any 720 or 360 floppy disks to burn the image on. So, if anybody is running this DOS, how does FreeBasic work on this setup. For that matter, how did/does the multitasking part enhance the DOS. I wonder why this version never really took off.

With the possible availability of the Night Kernel for FreeDOS, I wondering what will different from the original MS-DOS 4.0 version. Just very curious.

[rugxulo]

I've never used it. I'm not sure how legal it is to download or use these days (at least in U.S.).

IIRC, it's just a precursor to OS/2 1.x (16-bit), thus it also used NE format .EXEs. Not sure it will multitask the other stuff. It might also be real-mode only (whereas OS/2 could do 286 pmode stuff, even Win 3.x could do cooperative multitasking in standard/286 mode with its non-OS/2-compatible NE/Win16 .EXEs).

DR-DOS 7.03 is still sold online ($79 USD), and it has preemptive multitasking on 386s (but 64 MB per task maximum, requires its own proprietary EMM386 + DPMI + TASKMGR loaded). That would "probably" work with most FBC stuff.

You could also take a look at PC-MOS, but I wouldn't get my hopes up very high for much compatibility there. (Maybe I'm wrong, never bothered trying it.)

There are maybe other options, but I'm not sure about the availability/legality of them. (Desqview/X or whatnot.) Oh, you can still buy OS/2 (Arca Noae).

Honestly, even though imperfect, you may just wish to use Linux + DOSEMU2 (or slightly older DOSEMU). Or something like QEMU + FreeDOS atop FreeBSD. Who knows what would float your boat.

[dasyar]

Well, it looks like FreeDOS is the only one left that has some support, and sources that are available. I checked out DR-DOS, and that WEB site is no longer available, I guess they are no longer doing any business, with DOS. In fact, EDR-DOS OpenDR-DOS is no longer active.

Pat Villani, the author of FreeDOS is no longer available. I believe he had a commercial multitasking DOS, but the source code for that is nowhere to be found.

I know that there is very little activity or interest around DOS, at this time, but it seems to me that if it were updated, maybe it could pick up some interest again. Or maybe not. I am interested in the simplicity of DOS, but I am also interested in the GUI aspect of an OS. I wonder how complicated DOS would become if, not only if it had multitasking, but it also had multiuser capabilities.

[marcov]

I wonder how complicated DOS would become if, not only if it had multitasking, but it also had multiuser capabilities.

It would simply no longer be dos, but a successor OS. This already happened partially, the result was called "win9x"

[badidea]

I wonder how complicated DOS would become if, not only if it had multitasking, but it also had multiuser capabilities.

It would simply no longer be dos, but a successor OS. This already happened partially, the result was called "win9x"

I thought that Win9x was mostly a graphical shell on top of DOS. Not until WinNT did Windows become a real multi-user OS.

[marcov]

I wonder how complicated DOS would become if, not only if it had multitasking, but it also had multiuser capabilities.

It would simply no longer be dos, but a successor OS. This already happened partially, the result was called "win9x"

I thought that Win9x was mostly a graphical shell on top of DOS. Not until WinNT did Windows become a real multi-user OS.

Well, calling win9x multiuser is maybe indeed a bridge too far. But it was a whole lot more than a graphical shell. It had own drivers for various hardware and heaps and heaps of application API.

The point was that if you only allow extremely well behaved dos apps for your next gen "dos" kernel, you might as well use Linux.

[dasyar]

Just to clarify, I think of multiuser, in the "classic" sense where you have a port available for somebody to login and run a session. This means that you could have more than one person that is using the system, at the same time. Just think of the original Unix system.

Now, the way I recall it, back in the day, first you had DOS, then along came Windows 3.0. Basically you had to have DOS installed in order to install Windows 3.0, usually it was considered an upgrade. Then came along Win 9x, which most people got as an upgrade, which means you had to have DOS and Windows 3.0 installed before you could install Win 9x. I believe Win 9x was offered as a stand alone CD, at some point, at a much costlier price point.

Everybody keeps mentioning Linux, well I have tried most of the distributions, and most are not as simple as DOS to work in. I would now say that most of the popular Linux distributions are as bloated as Windows xx.

[marcov]

Just to clarify, I think of multiuser, in the "classic" sense where you have a port available for somebody to login and run a session. This means that you could have more than one person that is using the system, at the same time. Just think of the original Unix system.

Yeah. Initially I read it as having profiles, but the Unix definition came later. But in Unix init.rd scripts this had special meaning beyond all this.
But while multiuser is good for a server OS, I'm not 100% sure it is an absolute requirement for a desktop OS.

Unix and NT both come from the server side, so are bad examples in this regard.

Now, the way I recall it, back in the day, first you had DOS, then along came Windows 3.0. Basically you had to have DOS installed in order to install Windows 3.0, usually it was considered an upgrade.

Afaik WFW (windows 3.11) came with an installer that also installed the rudimentary Dos.

Then came along Win 9x, which most people got as an upgrade, which means you had to have DOS and Windows 3.0 installed before you could install Win 9x. I believe Win 9x was offered as a stand alone CD, at some point, at a much costlier price point.

Yes, but what has offering upgrades options to do with if an OS is separate or not? One might as well say that Linux is Dos too, since scores of people booted it using loadlin in the nineties to maintain a dual boot.

Everybody keeps mentioning Linux, well I have tried most of the distributions, and most are not as simple as DOS to work in. I would now say that most of the popular Linux distributions are as bloated as Windows xx.

"bloat" is a word that only conveys sentiment. It has no defined meaning. Ever considered that learning something new is simply hard? Learning dos from start would be hard for people with no exposure to it either. Specially since nearly every application has its own way of configuring/installing etc.

[caseih]

But while multiuser is good for a server OS, I'm not 100% sure it is an absolute requirement for a desktop OS.

Unix and NT both come from the server side, so are bad examples in this regard.

I'm not sure about that. Unix was originally an evolution/competitor of the interactive, time-slice multiuser OS Multics. It's primary purpose was to provide real-time interactive computing to multiple users at the same time, through attached terminals. There really weren't such things as "servers" back then. they were all "desktops" really. VMS (and NT) were in a similar vein.

The line between multi-process and multi-user is pretty blurry. The same mechanisms that protect and isolate processes also provide multi users. And most systems at the very least should offer protection system to protect the kernel from the user processes. Particularly when your user process is processing data from very unsafe places, like the world-wide web!

I guess I'm saying that multi-user capabilities works quite well to provide some safety and security in our modern world, even if only one person is using the computer most of the time.

[marcov]

But while multiuser is good for a server OS, I'm not 100% sure it is an absolute requirement for a desktop OS.

Unix and NT both come from the server side, so are bad examples in this regard.

I'm not sure about that. Unix was originally an evolution/competitor of the interactive, time-slice multiuser OS Multics.

Afaik the original one was for telephone switchboard control. Then later when they moved to PD11 they started added features, among others from multics.

And whatever you consider the righ history, it was not desktop (single person personal computer) oriented. It doesn't matter really if the clients are terminals or pcs.

The line between multi-process and multi-user is pretty blurry.

multitask or multiprocessing? Having multiple hardware threads of control is pretty advanced.

I was more thinking about e.g. AmigaOS, older BeOS etc.

The same mechanisms that protect and isolate processes also provide multi users.

Protection and multitasking is yet another thing. E.g. win3.11 or older classic MacOS did multitasking just fine (as long as you consider "cooperative" fine), but sucked at protection.

And most systems at the very least should offer protection system to protect the kernel from the user processes. Particularly when your user process is processing data from very unsafe places, like the world-wide web!

So the primary goal of an OS, is to access the WWW? How very, euh, Chome ! :-)

I guess I'm saying that multi-user capabilities works quite well to provide some safety and security in our modern world, even if only one person is using the computer most of the time.

Nowadays that nearly all CPus have hardware protection most of these distinctions are moot for current hardware. But you can't project such sentiments onto history, and this thread is called "msdos 4.0".

Even in modern IoT such distinctions still exist (e.g. rpi vs arduino, the common atmel version lack even the most basic CPU features). There are also non MMU MIPS (PIC32) and ARM (cortex M*) derivatives that lack complete hardware protection yet run e.g. FreeRTOS.

[rugxulo]

I checked out DR-DOS, and that WEB site is no longer available, I guess they are no longer doing any business, with DOS.

I'm surprised. That must be a new change (after many, many years). It was fine a few months ago! I doubt they'll stay offline for long, considering their history, but who knows. I bought a copy many years ago (2004?). It's quite ancient (circa 1999), fairly limited in many ways but still overall good. I was glad it could still be purchased online, at least for comparison. (The other major DOS clone still sold online is Datalight's ROM DOS, but that doesn't multitask, AFAIK, and I never bothered trying it.)

In fact, EDR-DOS OpenDR-DOS is no longer active.

Not active, but you can find what's left of it on WayBack. But it's not free/libre, rather "non-commercial use only" (kernel and shell only, nothing else, AFAIK).

Pat Villani, the author of FreeDOS is no longer available. I believe he had a commercial multitasking DOS, but the source code for that is nowhere to be found.

That initial concept wasn't quite fully DOS compatible, more of a portable "int 21h" emulator that supposedly barely worked well for what tools and uses were needed. It had to run atop 68k cpus, apparently. So it's not saying much that it multitasked. (And yes, Pat died in 2011, RIP.)

DOSEMU2 can (optionally) use the FreeDOS kernel, so that's a more obvious solution for multitasking, IMHO.

I know that there is very little activity or interest around DOS, at this time, but it seems to me that if it were updated, maybe it could pick up some interest again. Or maybe not. I am interested in the simplicity of DOS, but I am also interested in the GUI aspect of an OS. I wonder how complicated DOS would become if, not only if it had multitasking, but it also had multiuser capabilities.

It won't attract much interest, no, but it could raise a few eyebrows. But it's already got a few geniuses (not me) interested in it. Certainly I don't expect any miracles, but FreeDOS is already quite good, IMHO. Though not everybody uses FreeDOS, but that's where some of the important activity is happening (albeit mostly userland). DOS is a very dysfunctional family (many clones, many compilers, many different uses).

[caseih]

So the primary goal of an OS, is to access the WWW? How very, euh, Chome ! :-)

Pretty much! Gets even worse when you add in IoT ideas that everything should be connected just because. At that point, even programs running on single-tasking, bare-metal chips like Atmel can be vulnerable to outside attack. What an exciting world.

[caseih]

I know that there is very little activity or interest around DOS, at this time, but it seems to me that if it were updated, maybe it could pick up some interest again. Or maybe not. I am interested in the simplicity of DOS, but I am also interested in the GUI aspect of an OS. I wonder how complicated DOS would become if, not only if it had multitasking, but it also had multiuser capabilities.

DOS means lots of different things, though. Which aspect of DOS are you pining for? You say simplicity, but what does that really mean? I agree with you that I remember the simplicity of DOS with fondness also, but that simplicity has a specific meaning for me. For you what kind of simplicity are you wanting to have again? Are you talking about the interrupt-driven API the BIOS and MS-DOS provided you? The ease of developing in assembly language because of that API? The power of text-mode applications?

My first experiences with DOS go back to PC-DOS 1 days. And I remember the boom years of the MS-DOS software industry. Like many of you I cut my programming teeth with the likes of DOS interpreters, compilers, and IDEs such as BASICA, GW-BASIC, QuickBasic, TurboBasic, PowerBasic, Turbo C++, Turbo Pascal, etc. In those environments, I really enjoyed how fast and easy it was to write programs that did interesting things, albeit usually in text mode or with rudimentary graphics. I never did transition to Windows API programming. So for me the nostalgia for the simple days of MS-DOS refers to this this programming experience specifically, which I found again in Linux--Python now fills the same niche as BASIC used to for what I need to do. Although I now do develop more complicated graphical apps for Windows using tools like Qt and C++. With experience Windows isn't quite as overwhelming a programming target as it once was. But I still prefer writing text-mode apps.

As for simplicity, I recognize now that programming large and useful apps back in MS-DOS days, such as word processors, spreadsheets, and other commercial application packages was most certainly not simple by any definition. In fact it was incredibly complicated by today's standards. No one used BIOS screen writing routines because they were too slow. Everyone wrote directly to video memory, which meant they had to detect what video system they were running under. Printing? Oh boy. Graphics? Well we had Hercules, CGA, MCGA, Tandy, most of which didn't have any real BIOS support. MS-DOS and the BIOS were really insufficient early on for the needs of nearly all software on the market. We can remember those days with fondness, but let's not kid ourselves! Remember all the articles in the 80s about overcoming DOS limitations? MS-DOS apps back then had to basically be their own OS. I argue that even FreeBASIC apps, when compiled on MS-DOS, provide their own mini OS compared to what MS-DOS offers it.

Ever played with Geoworks Ensemble? Amazing environment/mini OS that ran on top of MS-DOS in 16-bit real-mode.

[paul doe]

... We can remember those days with fondness, ...

Ah, yes... the days of segmented memory and banked VESA... how could one forget those endless hours of pure fun? ;)

[angros47]

Ever played with Geoworks Ensemble? Amazing environment/mini OS that ran on top of MS-DOS in 16-bit real-mode.

Yes, it is now under Apache License, free and open source: https://github.com/bluewaysw/