OSDev.org

Hey guys,

There's a new operating system that I'm working on that targets the i8086 and newer hosted at https://cos.candlhat.org/. You may wonder why I'm creating a new operating system and the answer is that MS-DOS/PC-DOS has their limitations (i.e. no command history) and if you get FreeDOS from https://freedos.org none of the images seem to work with the i8086. There are places you can get FreeDOS for 8086 (but you don't exactly know exactly what's in the images) or you can get the source code and compile it yourself (provided you know how to and have the specific tools). My operating system is built with 8086 in mind so no sketchy downloads and no recompiling from source code. There's no internet support in the OS so there's no way of data being used so there's no need to even collect any data in the first place. The source code isn't available yet as I need to make a living and I'm testing out a pay what you want model first and if it seems viable I'll release the source code. If you find any bugs please feel free to post about them here (with details of how you got them) and I'll do my best to address them.

I don't really understand. If I buy your binary image, I know about as much about what's in the image as I do with MS-DOS, or with the 8086 FreeDOS images you spoke about. Only if I compile from source can I be somewhat confident in the result (the Ken Thompson attack notwithstanding).

robertapengelly wrote: ↑Wed Jul 31, 2024 10:45 am The source code isn't available yet as I need to make a living

I doubt you will make enough money from this venture to support a life. There are very few 8086 enthusiasts in the world, and even fewer that would consider paying for the OS. And even if you manage to sell to all of them, you can only do so once, but you need food constantly.

> I know about as much about what's in the image as I do with MS-DOS, or with the 8086 FreeDOS images you spoke about. Only if I compile from source can I be somewhat confident in the result

I understand that. You'd be trust some random guy on the internet with your data. MS-DOS was closed source so you had to trust Microsoft, back then I don't think it was an issue but nowadays they collect god knows what about you but yet people still buy Windows (mainly cause they need certain software). Compiling from source is fine provided you know how to and by all means people can choose to sit there and compile FreeDOS and some early versions of MS-DOS that were released on github, that's fine, the project I'm working on is a hobby and if I can make a little money from it then that'll be great.

> I doubt you will make enough money from this venture to support a life.

It's not about the amount of money really, I'm just trying to make some as well as see if there are actually people interested in the project. Plus the whole pay what you want model is kinda like taking donations, people donate to ReactOS and FreeDOS. ReactOS is a mess and really buggy even after all these years and FreeDOS is really good and by all means use that if you'd prefer.

> https://en.wikipedia.org/wiki/DOSKEY

I didn't know about that. Looking at the page I see ""DOSKEY was introduced in MS-DOS/PC DOS 5.0" but neither of them seem to work in pcem with an 8086 CPU.

I'm not clear what your aims are.

1. To produce an MS-DOS clone that will run existing software in an emulator, or
2. To produce a new OS that will run on 8086/8088 hardware.

In other words, are you concerned with running existing MS-DOS software or just with supporting old hardware?

If 1, I don't see that it matters whether a particular OS (e.g. FreeDOS) will run in a particular VM; just use a different VM and/or a different OS. Are you sure that you can account for all the undocumented features of DOS that some software uses? Does it matter if, for example, FreeDOS doesn't work with an 8086? Just emulate a processor that it does work with.

If 2, what software are you going to provide to run on your OS? (An OS without software is pretty useless.) Will you produce the equivalents of WordPerfect, VisiCalc, dBase, Turbo Pascal, Turbo C/C++, etc. And are you going to document your API to allow others to create programs for your OS?

I'm not sure that the world needs another DOS clone, but I think the task of producing a new OS with a comprehensive suite of software is a mammoth undertaking. If the aim is to provide some missing features of MS-DOS (you mention command history), would it not be more realistic to produce add-ons (a la DosKey) that provide this functionality to an existing installation?

I wish you good luck, but I'm not convinced that there is much demand for either 1 or 2; not to the extent that people will pay for it when there is so much excellent, open-source software available.

If I understand correctly, the project is aiming for binary compatibility with MS-DOS, targeting 8086 hardware, but with some quality of life features which are currently only available in other projects that don't supply compiled distros which support 8086.

It's a niche within a niche within a niche and I think it'll struggle to find a user base - the only users I can envisage are retro computing enthusiasts with fairly rare machines that they want to actually run software on regularly, but despite using period accurate hardware and software they wouldn't also insist on a period accurate OS, also they don't know how to compile FreeDOS from source.

As a hobbyist project I can see the appeal though. It has a well defined scope, it isn't so ambitious as to be insurmountable and it'd be interesting from a computer/software history angle.

Right first part is for iansjack (I don't know how to or if you can tag people on here).

> I'm not sure that the world needs another DOS clone.
> when there is so much excellent, open-source software available.

Using the second sentence mainly, the same can be said about other projects. Why does ReactOS need to exist when Windows NT based operating systems exists? Why do we need so many flavors of Linux? Why do we need so many Linux distros (Debian, Ubuntu, Linux Mint, PopOS, Fedora, etc) when under the hood they are still Linux after all? Talking about distors, why do we need Ubuntu, Linux Mint, PopOS, etc when Debian exists given how they are using the same (or similar) code base? The same can be applied to Linux Mint, PopOS, etc when Ubuntu exists as they are using the same (or similar) code base? The list goes on. Each project targets a different spin on existing projects, not sure what Debian uses but Ubuntu uses GNOME as the DE (they tried Unity at one point but abandoned it), Linux Mint uses the Cinnamon, and so on. There's a lot more to each project than just the DE but that's just an example. The same goes for my project, it's targeting 8086 and I'm hoping to eventually get it fully MS-DOS compatible. Why MS-DOS? The answer is simply the FAT filesystems. FAT is well documented and supported by practically all OS's these days, you can simply use modern systems to transfer files to a disk and have them available in my OS. Not only is the filesystems a factor but the MS-DOS API is well documented as well meaning that it's easier to implement.

Now StudlyCaps (again I don't know how to or if you can tag people on here).

> The only users I can envisage are retro computing enthusiasts with fairly rare machines.

That's exactly who I'm targeting. I have a fascination with vintage systems. Things were much simpler back then. Nowadays you need hundreds of megabytes of ram (sometimes even gigabytes) to even run an OS. I know technically some can run on less than whats expected (e.g. Windows) but you risk performance issues.

> Also they don't know how to compile FreeDOS from source.

It's not always about people not knowing how but maybe people haven't got the time or patience to compile from source. I've never compiled FreeDOS as I've never got round to setting up the development environment but let's say each time it takes 15 - 30 minutes to compile and you lose the files that's another 15 - 30 minutes to recompile again, whereas my project is aimed at vintage machines so it's just a quick download and your done (depending on your internet).

robertapengelly wrote: ↑Wed Jul 31, 2024 8:11 pmLooking at the page I see ""DOSKEY was introduced in MS-DOS/PC DOS 5.0" but neither of them seem to work in pcem with an 8086 CPU.

That sounds like a problem with pcem (or how you're configuring pcem). MS-DOS 6.22 works perfectly fine on an IBM PC 5150, as long as you install enough RAM upgrades. DOS 5 should work just as well.

robertapengelly wrote: ↑Thu Aug 01, 2024 9:33 am(I don't know how to or if you can tag people on here)

There's a "quote" button in the corner of every post, you can click it to add a quote box that will nicely format the text you're replying to and tag that person. You can quote multiple posts at once by scrolling down below the post edit box and clicking the "quote" button on each post you want to quote.

Octocontrabass wrote: ↑Thu Aug 01, 2024 11:04 am That sounds like a problem with pcem (or how you're configuring pcem).

I think it's a ROM issue, I tried it again with 3 different ones this time (one was 8088) and it works on 2 of them. If not a ROM issue then it's a sit and wait game due to how slow older machines are and I don't have the patience to wait (big flaw of mine).

Octocontrabass wrote: ↑Thu Aug 01, 2024 11:04 am There's a "quote" button in the corner of every post, you can click it to add a quote box

Thank you.

I think it's a ROM issue though cause I've had it running while writing this and still no luck.

Based on feedback I've decided to make the project open source regardless. It can be found at https://git.candlhat.org/chimaera.git so have fun examining it.