DOS ain't dead

Just found this from a chat room:

https://github.com/buricco/doslite
"DOSLITE Project - Lightweight attempt at rewriting DOS 6.21."

I wonder if a proper lightweight COMMAND.COM will be available in the future.

Hmm...

> If the reference to software freedom makes DOSLITE sound a lot like FreeDOS,
> it is not entirely a coincidence. Both operating systems are MS-DOS clones,
> and both operating systems aim to maintain software freedom, but the
> similarity ends there: where FreeDOS is the "GNU" of MS-DOS clones, DOSLITE
> aims to be the "BSD". As one example, FreeDOS applications are often bigger
> but more powerful versions of those from MS-DOS, while their DOSLITE
> counterparts aim to be exactly as powerful as those of MS-DOS and, often,
> smaller. "Codegolfing", or the process of whittling away code to produce the
> smallest result, is not uncommon.

---
https://gitlab.com/tkchia · https://codeberg.org/tkchia · 😴 "MOV AX,0D500H+CMOS_REG_D+NMI"

Hmm...

given that more than a dozen (14 to be exact) programs are complete enough to have a binary

another almost a dozen (11 to be exact) are described as 'Here lie several unfinished implementations of MS-DOS utilities. Some of them are barely stubs, and some of them are semi-functional. Many of them are very buggy.'

https://github.com/buricco/doslite/blob/main/exper/readme.txt




and command.a86 contains a section
; A list of commands that would really need to be implemented into COMMAND.COM
; proper once the framework is built for it:
;
; CALL              (modifies the batch state machine)
; EXIT              (exit from what?)
; FOR               (relies heavily on the internal COMMAND.COM infrastructure
;                    which isn't finished yet)
; GOTO              (modifies the batch state machine)
; IF                (needs access to errorlevels)
; PATH              (modifies the environment)
; PROMPT            (modifies the environment)
; SET               (modifies the environment)
; SHIFT             (modifies the batch state machine)
;

and no kernel in sight: this isn't going anywhere

> Just found this from a chat room:
>
> https://github.com/buricco/doslite
> "DOSLITE Project - Lightweight attempt at rewriting DOS 6.21."
>
> I wonder if a proper lightweight COMMAND.COM will be available in the
> future.

"DOSLITE aims to replace components of MS-DOS piece by piece,
eventually to include the kernel."

IMO...

Making a 100% functional kernel along with a 100% functional command.com
would be the necessary first step after-which each of the 'components'
can then be re-written to work with the new kernel & command.com

---
--
http://glennmcc.org/

> Just found this from a chat room:
>
> https://github.com/buricco/doslite
> "DOSLITE Project - Lightweight attempt at rewriting DOS 6.21."
>
> I wonder if a proper lightweight COMMAND.COM will be available in the
> future.

Their Debug and GRAFTABL was discussed previously on the freedos-devel mailing list, though we didn't get any source revision of their patches to Debug at the time.

---
l

I released what I had knowing that it was far from ready for primetime. I'm not exactly the best programmer on earth and my ASM-fu is probably about on par with my C-fu when I tried coding a DOS clone previously. I've gotten quite a bit further this time around than I did 20 years ago.

I did actually implement a fairly complete COMMAND.COM back then but it has some serious issues.

Uploading to Github was more to get suggestions and advice to try to improve it, and I kind-of wanted to wait until I got further. Particularly, I was trying to rewrite Robert Nordier's FORMAT.COM into ASM, but that code is a total rat's nest. There's a lot of half-finished stuff in the "exper" folder because I'm having trouble figuring out how to implement.

I think a lot of DOS 6.2 could be optimized a bit, and run fine on a system with 3.31, which would make it more useful on a 286. The kernel is actually lower priority since I can just temporarily host the userland on an MS-DOS kernel - and because one of the things I'm also trying to do is use my code to torture test other kernels (like RxDOS). Besides - the userland is enough of a challenge.

I did leave some notes as to how perhaps the MS-DOS 2.11 command.com could be brought up to what is needed for a DOS 6.2 clone. I'm not sure I know enough to actually do the implementation though.

Divide and conquer is an important part of my strategy here. If something has DOS 2.11-level support - go up one version at a time until reaching 6.21 (this was the first strategy I used with FIND, for example, until I just got annoyed trying to read Microsoft's code and just knuckled down and rewrote it from scratch).

https://6.buric.co/rmfcom63.zip

This is the COMMAND.COM I wrote 20 years ago, for what it's worth. It's extremely buggy and fragile, and definitely the work of a C n00b who was used to working in BASIC.

> https://6.buric.co/rmfcom63.zip
>
> This is the COMMAND.COM I wrote 20 years ago, for what it's worth. It's
> extremely buggy and fragile, and definitely the work of a C n00b who was
> used to working in BASIC.

is there ANY, like in ANY ANY, reason not to RIP?

> > https://6.buric.co/rmfcom63.zip
> >
> > This is the COMMAND.COM I wrote 20 years ago, for what it's worth. It's
> > extremely buggy and fragile, and definitely the work of a C n00b who was
> > used to working in BASIC.
>
> is there ANY, like in ANY ANY, reason not to RIP?

There is this little thing. It's called emotions. Humans have these.

---
Forum admin

> > > https://6.buric.co/rmfcom63.zip
> > >
> > > This is the COMMAND.COM I wrote 20 years ago, for what it's worth.
> It's
> > > extremely buggy and fragile, and definitely the work of a C n00b who
> was
> > > used to working in BASIC.
> >
> > is there ANY, like in ANY ANY, reason not to RIP?
>
> There is this little thing. It's called emotions. Humans have these.
>

Yep...
same reason why I kept working on DOS Arachne till it has now
reached the point of the absolute limit of my programming skills
instead of allowing it to RIP 10 years ago.

---
--
http://glennmcc.org/

> There is this little thing. It's called emotions. Humans have these.
>

And nostalgia. I will never let DOS or my DOS projects RIP. It is just too much fun to keep tinkering away. Nothing better than getting back into a project or a program after 20 years of inactivity. It makes me *stronger* and yes, emotional to think I've still got it me, still got the knack. Can't let it die.

> > There is this little thing. It's called emotions. Humans have
> these.
> >
>
> And nostalgia. I will never let DOS or my DOS projects RIP. It is just too
> much fun to keep tinkering away. Nothing better than getting back into a
> project or a program after 20 years of inactivity. It makes me *stronger*
> and yes, emotional to think I've still got it me, still got the knack.
> Can't let it die.

Part of that too.

There's two different mindsets here, and both are understandable. One side says if it's not perfect or close to it, why bother? (It's like my Detective Conan fansubs: I often had to edit existing translations or cross-translate from Catalan rather than translate directly from Japanese. But the results, while not perfect, were still an improvement.) Another side says, there's always going to be someone interested.

I had actually been working on the relaunch for some time. I wanted to wait until it was obvious I wasn't just messing around before releasing widely, and I went with GitHub to try to encourage external contribution. (Which worked; I got some improvements for DEBUG from ecm. I also got some improvements to GRAFTABL from her because I shared it on the FreeDOS list - although it is a nearly useless utility, which I'll readily admit, it has its purposes, was easy to throw together, and could replace an unfinished version on the FreeDOS site.)

(ETA: and if I haven't already said it, ecm has helped me improve my 8086-fu. )

> This is the COMMAND.COM I wrote 20 years ago, for what it's worth. It's
> extremely buggy and fragile, and definitely the work of a C n00b who was
> used to working in BASIC.

I am also, have a pet project - "command.com", based on code "Small Basic" by Herbert Schildt in C. Now, it is allow to run .bat files under linux. But lacks some modern features like "FOR" and redirection (i dont remember).

So, i don't publish it (because it stupid project - who need it?).

If it need for somebody, possible continue it and extend, and make port to Sphinx C-- (to have extra small size).

---
Make DOS great again!

Carthago delenda est, Ceterum censeo Carthaginem delendam esse.

> I had actually been working on the relaunch for some time. I wanted to
> wait until it was obvious I wasn't just messing around before releasing
> widely, and I went with GitHub to try to encourage external contribution.
> (Which worked; I got some improvements for DEBUG from ecm. I also got some
> improvements to GRAFTABL from her because I shared it on the FreeDOS list -
> although it is a nearly useless utility, which I'll readily admit, it has
> its purposes, was easy to throw together, and could replace an unfinished
> version on the FreeDOS site.)
>
> (ETA: and if I haven't already said it, ecm has helped me improve my
> 8086-fu. )

Any news on DOSLITE?

There has been not progress on GitHub or https://6.buric.co/doslite/ for the last two years.

---
Forum admin

> > I had actually been working on the relaunch for some time. I wanted to
> > wait until it was obvious I wasn't just messing around before releasing
> > widely, and I went with GitHub to try to encourage external contribution.
>
> > (Which worked; I got some improvements for DEBUG from ecm. I also got
> some
> > improvements to GRAFTABL from her because I shared it on the FreeDOS list
> -
> > although it is a nearly useless utility, which I'll readily admit, it
> has
> > its purposes, was easy to throw together, and could replace an
> unfinished
> > version on the FreeDOS site.)
> >
> > (ETA: and if I haven't already said it, ecm has helped me improve my
> > 8086-fu. )
>
> Any news on DOSLITE?
>
> There has been not progress on GitHub or https://6.buric.co/doslite/ for
> the last two years.

no idea, but now we have a light-weight command.com, i.e. SvarCOM.

> Any news on DOSLITE?
>
> There has been not progress on GitHub or https://6.buric.co/doslite/ for
> the last two years.

Yeah, haven't been able to write anything more for it lately.

-uso.