DOS ain't dead

I'm looking for a simple file manager written in C, opensource, that I can modify to adapt it to the serial console of SBC188, can someone tell me where I can download it? thank you

> I'm looking for a simple file manager written in C, opensource, that I can
> modify to adapt it to the serial console of SBC188, can someone tell me
> where I can download it? thank you

Gram's Commander might be a good candidate....

https://github.com/gramster/gc3

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

> > I'm looking for a simple file manager written in C, opensource, that I
> can
> > modify to adapt it to the serial console of SBC188, can someone tell me
> > where I can download it? thank you
>
> Gram's Commander might be a good candidate....
>
> https://github.com/gramster/gc3


wow, nice as soon as I have time I try to get my hands on it, I use borland turbo c as a compiler, do you think it could be compatible? or do you recommend another compiler?

There's mentioned
"gc3 compiles and runs under MS-DOS with Borland or Turbo C/C++ v3.x."
but I'm unable to correctly generate the secondary makefilke. Not sure what version of make should I use. Should I configure it under linux and then switch to dos?

BTW some years ago I started to write my own file manager in DJGPP but it was left very unfinished.

BTW2 I see you have a nice toy - SBC188. Did you considered 86Duino? It's not so retro but much powerfull and still running DOS and use opensource coreboot instead BIOS. It also doesn't have VGA by default but I was able to make simple adapter to connect with common PCIe VGA by single x1 lane.

---
DOS gives me freedom to unlimited HW access.

Hello RayeR,

> There's mentioned
> "gc3 compiles and runs under MS-DOS with Borland or Turbo C/C++ v3.x."
> but I'm unable to correctly generate the secondary makefile. Not sure what
> version of make should I use. Should I configure it under linux and then
> switch to dos?

Yeah, the rule for generating Makefile.2 in the main makefile seems to be borked...

Makefile.2: site.def Makefile.gen
        -type site.def > Makefile.2
        -type Makefile.gen >> Makefile.2
        -cat site.def Makefile.gen > Makefile.2

The first two commands are correct for MS-DOS, but unfortunately the third command basically destroys their good work and just leaves a "Bad command or file name" (or some such) in Makefile.2. Maybe for now, just manually do

type site.def > Makefile.2
type Makefile.gen >> Makefile.2

(under DOS) then carry on with the

make dos

?

Thank you!

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

> There's mentioned
> "gc3 compiles and runs under MS-DOS with Borland or Turbo C/C++ v3.x."
> but I'm unable to correctly generate the secondary makefilke. Not sure what
> version of make should I use. Should I configure it under linux and then
> switch to dos?


I'm using turbo c v2.04, i used dosbox to create my ansi vt100 compatible text editor, now i can compile things directly from inside SBC188.

>
> BTW some years ago I started to write my own file manager in DJGPP but it
> was left very unfinished.
>
> BTW2 I see you have a nice toy - SBC188. Did you considered 86Duino? It's
> not so retro but much powerfull and still running DOS and use opensource
> coreboot instead BIOS. It also doesn't have VGA by default but I was able
> to make simple adapter to connect with common PCIe VGA by single x1 lane.

i like vintage things, to build my SBC188 i also found an intel original i80188 with ceramic package

Nice board, I never seed 186/188 in ceramic pkg before, I have only PLCCs...
Maybe soneone would rather build 386 SBC, here's one
http://www.alexandrugroza.ro/microelectronics/system-design/isa-80386dx-sbmc/index.html

---
DOS gives me freedom to unlimited HW access.

> Nice board, I never seed 186/188 in ceramic pkg before, I have only
> PLCCs...
> Maybe soneone would rather build 386 SBC, here's one
> http://www.alexandrugroza.ro/microelectronics/system-design/isa-80386dx-sbmc/index.html

https://www.cpu-world.com/CPUs/80188/Intel-A80C188XL20.html

OK, I got a step further, Makefile.2 created but it end with bunch of errors in gc3.h and some other headers (BCC 4.52). I also tried to compile linux version under Debian Linux but had no luck too.
http://www.rayer.g6.cz/1tmp/g3c.txt
Could you upload compiled DOS binary?

---
DOS gives me freedom to unlimited HW access.

> https://www.cpu-world.com/CPUs/80188/Intel-A80C188XL20.html

20 Mhz? Interesting.

I personally would compare 8086 and 186 output from various compilers, out of curiosity, e.g. Oberon-M 1.2. (186 ENTER/LEAVE is actually four times slower on some of my semi-modern machines! Well, mostly ENTER.)

Have you seen that old Tandy ad with Bill Gates? You know, the Tandy 2000 used a 186, also. While I don't remember exactly, LGR on YouTube made various Tandy videos.

> > https://www.cpu-world.com/CPUs/80188/Intel-A80C188XL20.html
>
> 20 Mhz? Interesting.
>
> I personally would compare 8086 and 186 output from various compilers, out
> of curiosity, e.g. Oberon-M
> 1.2. (186 ENTER/LEAVE is actually four times slower on some of my
> semi-modern machines! Well, mostly ENTER.)
>
> Have you seen that old
> Tandy
> ad with Bill Gates? You know, the
> Tandy 2000 used a 186,
> also. While I don't remember exactly, LGR on YouTube made various
> Tandy videos.


tandy 2000 cost less and performed better than IBM AT and early 286


186 and 188 have the same instruction set as the 286, except the protected mode instructions. I have read from many places that it is faster than 286 at the same clock thanks to the integrated peripherals.

The 186 architecture was better, the idea of IBM and Intel was that software houses had to use BIOS to access the hardware, in this way the differences between 8086 and 186 disappeared, but many software houses made programs that directly access the hardware to get a little more performance, so 186 was not 100% compatible with 8086 and some programs didn't work well.

So it was bound to industrial controllers (produced until 2007) and a few commercial PCs in the 1980s.

> Have you seen that old
> Tandy
> ad with Bill Gates? You know, the
> Tandy 2000 used a 186,
> also. While I don't remember exactly, LGR on YouTube made various
> Tandy videos.

There was also the Tandy 1100 FD/HD, a laptop which had 640 KiB of RAM and a NEC V20 running at 8 MHz apparently. I have an old Nixdorf 8810, which is almost entirely the same as a Tandy 1100 FD. Unfortunately the diskette drive (90mm 720 KiB only) is not functional and I haven't gotten around to fix it yet. (And the power supply cable is unreliable, and of course the battery is long dead.)

The NEC V20 supports 186 level instructions, though it needs special detection for this model as opposed to the typical 186 check. The typical check is shift count masking with 31, 8086/8088 and the NECs do not mask large counts in cl, the Intel 186/188 and up do mask the counts.

---
l

> The NEC V20 supports 186 level instructions, though it needs
> special
> detection for this model as opposed to the typical 186 check. The
> typical check is shift count masking with 31, 8086/8088 and the NECs do not
> mask large counts in cl, the Intel 186/188 and up do mask the counts.

Jim Leonard (Trixter) once sent me some detection code for old cpus, but I can't remember where it is. I'll have to look for it. EDIT: Apparently here.

Anyways, HelpPC also had some old cpu detection info.

There were various old (buggy?) DOS detection tools, too, sometimes with sources. One was by Eric Auer (CPULEVEL). I remember nagging him to support 186 and 486 separately (instead of mistreating them as glorified 8086 or 386), pointing to some other obscure detection code that I can't remember at the moment.

Honestly, that kind of detection is somewhat tricky. I recommend making it possible to disable checking manually beforehand and/or override it manually via cmdline switch. (Testing is rarely exhaustive, and it's too easy to make mistakes.)

> Honestly, that kind of detection is somewhat tricky. I recommend making it
> possible to disable checking manually beforehand and/or override it
> manually via cmdline switch. (Testing is rarely exhaustive, and it's too
> easy to make mistakes.)

Good idea. The debugger uses its CPU detection for two purposes: Setting up the default machine type for the assembler and disassembler (can be overridden using the M0..M6 commands once at the command line), and determining whether to use the patches for 386+ operation or for non-386 operation during initialisation. (For lDebug here and FreeDOS Debug here. lDebug expanded these patch tables a lot.) Because the /C command line switch of lDebug only inserts into the command line buffer, which is read by the normal command line processing loop, it can't be used to override the init.asm decision on whether a 386 is present.

---
l

> Hello RayeR,
>
> > There's mentioned
> > "gc3 compiles and runs under MS-DOS with Borland or Turbo C/C++ v3.x."
> > but I'm unable to correctly generate the secondary makefile. Not sure
> what
> > version of make should I use. Should I configure it under linux and then
> > switch to dos?
>
> Yeah, the rule for generating Makefile.2 in the main makefile seems to be
> borked...
>
> Makefile.2: site.def Makefile.gen
> -type site.def > Makefile.2
> -type Makefile.gen >> Makefile.2
> -cat site.def Makefile.gen > Makefile.2
>
> The first two commands are correct for MS-DOS, but unfortunately the third
> command basically destroys their good work and just leaves a "Bad command
> or file name" (or some such) in Makefile.2. Maybe for now, just manually
> do
>
> type site.def > Makefile.2
> type Makefile.gen >> Makefile.2
>
> (under DOS) then carry on with the
>
> make dos
>
> ?
>
> Thank you!


I'm trying to compile gc3 with turboc 2.01 simply with the command tcc -1 gc3.c the compilation goes on for a while without errors, then reports the lack of 2 files that are not actually present "gcsite.h" and "direct.h "