Comment by nils-m-holm | Hacker News Reader

Comment by nils-m-holm | Hacker News Reader

Parent Full thread

If you have used a very early version of MSDOS, its is not very different. File names are 8+3 characters long with a dot in between, upper and lower case is not distinguished. There are only 5 "built-in" commands: DIR, ERA (erase files), REN (rename, but the syntax is REN NEW=OLD), SAVE, and TYPE.

There are no directories, but "user areas" numbered from 0 to 15, so different users can have different files on the same disk. User areas are selected with the USER command. However, most people used user areas as poor-man's directories.

There is no COPY command. Copying files (and much more) is done with a program named PIP (peripheral interchange program). For instance,

    PIP B:=A:*.COM[E]

would copy all files of the type COM from drive A: to drive B: and echo the file names.

What was great about CP/M was that it allowed programs to run on different computers, as long as they were 8080- or Z80-based. It was a thin, but very effective abstraction layer between the hardware and a program.

MSDOS, which was originally based on CP/M, was a step back, really, because everybody tried to program the hardware directly instead of going through the proper interfaces.

> MSDOS, which was originally based on CP/M, was a step back, really, because everybody tried to program the hardware directly instead of going through the proper interfaces.

That wasn't anything to do with the technical characteristics of MS-DOS versus CP/M. It was about the market conditions – for MS-DOS, the vast majority of users were running it on the IBM PC (and later compatible clones), so developers coded to the IBM PC BIOS and direct hardware access. With CP/M, hardware was far more diverse so developers had an incentive not to do that sort of thing.

It was also about speed. For example, going through BIOS for video was slooooow, so most everyone went straight to the hardware, which buggered up anything that wasn't an MDA/CGA/EGA/VGA video adapter or near identical (like those with a NEC 7220...there were one or two). It meant things like the Wyse 700 (high-res mono video adapter) had to have application specific drivers.

In theory, they should have written to MS-DOS API, and then if BIOS video is slow, Microsoft (or OEMs) could have changed the MS-DOS console device driver to use direct hardware access. The problem was, everyone just went straight to the hardware so there was little impetus on anyone to write a faster console device driver for MS-DOS.

Iirc the bios interface was inherently slow: output was a character at a time, one bios call each. It’s been a long time, though.

Not only that, but CP/M itself has a rather DEC-ish flavor to it. PIP was found on RT-11 and RSTS/E (maybe RSX-11 as well?), and the drive identifier with a colon is also a nod to DEC.

VMS feels like DOS if DOS were actually secretly UNIX

Also, don't forget about OS/8 for the PDP-8 series. Very DOS-y, and released by DEC all the way back in '71.

VMS feels like DOS if that DOS was from another dimension. About the only thing they have in common is "DIR" and ".EXE" extensions.

Which is also why it's so fascinating Windows NT sort of merged both parallel dimensions with its influences directly from both DOS and VMS.

It's not a secret that CP/M was influenced by RT-11. Lots of clear similarities. Kildall and his team used DEC machines as cross development machines.

And at the system level, all the MS-DOS system calls in version 1.0 were direct copies of system calls in CP/M. Made it very easy to translate CP/M programs to MS-DOS if you wanted.

Reply on news.ycombinator.com