50 years ago, CP/M started the microcomputer revolution Late in the summer of 1974, CP/M first started running on hardware. It became one of the first cross-platform microcomputer OSes, and revolutionized the hardware and software industries. You can now legally run the raw unbridled power of CP/M 2.2 anywhere you like You can now legally run the raw unbridled power of CP/M 2.2 …
Good article. In 1982, near the end of Digital Research and CP/M-80, a reasonable IBM PC cost somewhere in the £3000 to £4000 range.
.....but Adam Osborne was shipping the Osborne 01 (with CP/M development tools, Wordstar, SuperCalc, CBasic, MBasic) for £1500.
I got a three year loan from the bank and bought an Osborne 01. The Osborne and CP/M got me from idiot to competent in my spare time.
Buying dBase-II for CP/M in 1983 was also a huge help!!
Aside: The Osborne had 64K of main memory. The top 10K was where the CP/M OS resided. So all my programming work had to fit into around 50K.
If you look at the dBase-II memory image, dBase could not fit into 50K. So dBase had two blocks of memory: a core block in memory, and an overlay block which was swapped in and out elsewhere in memory.
Aside: if I can do word processing with Wordstar in a 64K memory machine, why am I using hundreds of megabytes for LibreOffice in 2024??
You yes, you only care about the words.
The marketing analyst writing a memo, the secretary writing a letter, or the person running a flyer for the church group do care about the font and positioning.
As a matter of fact, so many people care about font and position that, before the advent of WYSIWYG, LaTex was invented...
[Author here]
> before the advent of WYSIWYG, LaTex was invented...
You've got your timeline wildly out.
The first version of LaTeX was June 1984, about 6 months _after_ the first Apple Macintosh, three years after the Lisa and about a decade after Bravo on the Xerox Alto.
https://www.theregister.com/2023/03/16/the_xerox_alto_50_years/
So, not even close, no. TeX itself is a decade older, and contemporaneous with the Alto, but TeX is not even slightly WYSIWYG.
"TeX itself is a decade older, and contemporaneous with the Alto, but TeX is not even slightly WYSIWYG."
But they weren't saying that it was. They were saying that it was a non-WYSIWYG way of writing when you care about the visual design, rather than just writing down words, pointing out that, even without WYSIWYG software, people wanted to do that. Most of the LaTeX software isn't WYSIWYG either, though they've now got some that render instantly.
> But they weren't saying that it was. They were saying that it was a non-WYSIWYG way of writing when you care about the visual design
Hmmm. I have reread williamyf's post, and that's an interpretation, I suppose. I still think it's wrong:
* LaTeX is not TeX
* Neither LaTeX nor TeX is WYSIWYG
* TeX and WYSISYG are very loosely contemporaneous -- it's before WYSIWYG was common, certainly
But once we disentangle these things, we're left with "there were text-formatting layout tools before WYSIWYG". Well, yes, we know that. TeX wasn't the first by a _long_ shot
Early Unix had `roff` from ~1971. That was a Unix version of RUNOFF which goes back to ~1961 or so.
So the statement is empty, ISTM.
WYSIWYG came after non-WYSIWYG. Yes, we know that.
There was text formatting before WYSIWYG. Well obviously or showing it on screen wouldn't be a big deal.
There was style-based non-user-determined formatting before user-based formatting. Yes, that is what I'd expect, yes.
What is the core point here apart from making a wrong statement by getting a date a decade wrong?
1. First came offline rules-based formatting.
2. Then came user-defined formatting which wasn't rules based.
3. Then came tools that showed it to you as you did it.
4. After all of them came tools like LyX (1995), which is kinda sorta LaTeX but WYSIWYG.
And good luck to anyone wanting to do really complex layout such as splitting a map across facing pages using rules based layout. They're either going to have to be very, very good at applying the rules in their head or else use a lot of proofing runs to get good results. For that sort of writing using WYSIWYG is the equivalent of moving from punched cards and batch compiling to programming in an IDE.
"What is the core point here apart from making a wrong statement by getting a date a decade wrong?"
We agree that the dates are wrong. Their point, which is mostly irrelevant to TeX, how powerful it was, and when it existed, is that some people care about the visual layout. They said this in order to respond to someone who did not care about the layout:
AC: "if I was writing a modern replacement for "War and Peace", surely I'm interested in the words and not the fonts?"
Their main point was that people do care about layouts sometimes, and they are the reason why WYSIWYG tools existed, and before WYSIWYG tools were practical, non-WYSIWYG tools which handled more than just the words of a document were used. Incidentally, they never claimed that LaTeX was WYSIWYG as it was specifically their example of a non-WYSIWYG tool. They got the tool wrong in that description, but their core point, although it has mostly been lost to smaller details, is correct. Some people do care about more than the words. They are the reason why WYSIWYG editors have existed, and that is at least partially the reason why LibreOffice has so many features that take up so many resources.
In that case, you could just use Vim (Windows or Linux it's your choice).
And then pay someone to arrange it for printing (fonts, formatting and all that stuff).
A long time ago (yes, I'm grey bearded), I translated a book so I typed it using EDT on RSX-11 running on a local (East-European) flavour of PDP-7. Then printed everything on an Epson dot matrix printer.
Do you still want to go there ?
I wonder how many proof runs it took to get an acceptable layout. The original Unix development was effectively financed at Bell Labs as a means of WP of patent applications. A networking textbook is more hardcore than that.
OTOH try a book which has maps and photographs, keeping the text describing them onto the same or a facing page as far as possible. Add the complication of some of these being split across facing pages. That's harder core. And wondering why adding a few words at the op of a page has suddenly left a white gap at the bottom until you realise that there was no longer room for both a footnote an the paragraph that referenced it.
Complex layout is tricky. The advantage of WYSiWYG is that your proofing happens continuously on screen in front of your eyes.
I wrote my undergraduate thesis, which admittedly is a mere 47 pages single-spaced, in vi using roff. Wasn't difficult.
My master's thesis, actually a bit shorter at 77 pages but mostly double-spaced, I wrote in LyX, with some tweaking of the LaTeX in vi to comply with the exacting and arbitrary requirements of the Graduate School for depositing. Also not difficult.
Much of my academic writing in my first stint in graduate school I did in WordPerfect, which was OK. A number of those essays are longer than either thesis.
If I'd had to use Word, or anything running on a Mac, I'd have given up. I've used Word and Macs.
if I was writing a modern replacement for "War and Peace",
I think that would be rather a smaller market than the one for LibreOffice/Open Office/MSWord/Mac What ever they use/etc/
And pretty fonts do come in handy when you are making a sales leaflet, a menu, a letterhead etc.
... and you'd probably be using something like Scrivener, something that is designed to actually help write long books and scripts.
Scrivener is good at the organisation and construction of a long publication. Keeps track of it all.
It still goes into DTP for the layout. For me lately I go from scraps of text to Scrivener and from Scrivener to Affinity (or to Scribus if it will handle it).
if I was writing a modern replacement for "War and Peace", surely I'm interested in the words and not the fonts?
If you were writing it for self publishing you might well be more interested in fonts etc. Even more so if you were writing non fiction with a need to provide headings, sub headings, insert images and tables, keep track of references to these in the text etc. Alternatively you could use a plain text editor and then something like Scribus or InDesign to separate text from layout.
A word processor is more than a simple text editor. Both have their uses.
W&P written with a quill and bottle of ink. As were A Christmas carol, Pride and Prejudice, the 3 musketeers. More recent classics written with typewriters. easier in some ways. Remember a writer saying that his productivity dropped when he started to use a word processor because he kept editing and re-editing where with a typewriter he just decided enough was enough. Word processor just made it too easy to tweak things.
They may have had real pens with metal nibs for many of those stories, copper nibs possibly pre 79 ad.
Ditto, even earlier on a BBC micro. It came with a mouse too, since few people had even heard of a mouse back then :-)
AMX mouse with AMX Pagemaker (long before Aldus took the name for it's own DTP system)
I remember in the '90's I was writing library code in C and they wanted us to write documentation for each function as we went.
The project leader had set up a word perfect template in the text version of the WP. Simply by using the correct paragraph tags perfectly structured manual pages could be printed on the laser printer. I've never worked on such a well set up WP system before or since.
No WYSIWYG came along on the original IBM PC and that had a maximum of 1 megabyte RAM.
We did manage to produce useful software on 64k machines.
It used to be fun finding ways to make stuff quicker and smaller occasionally dropping into assembly language. When I finally retired it was all boring java (and the like) Lego programming (which relies on massive amounts of memory and fast processors) and efficiency is a dirty word (it costs money)
Libreoffice does a lot more. There's nothing stopping anyone using Wordstar on a modern system with a small amount of effort, or for that matter early versions of WordPerfect, but few people do.
Even at the time I was using both an Amstrad PCW and a 286 PC, the PC was far more functional. Supercalc just wasn't comparable to Lotus 1-2-3 at the time. WordPerfect did an awful lot more than Locoscript, Protext, or Wordstar.
CP/M is a good system, and I still have a few computers that can run it (PCW, Agon Console8, Spectrum Next, NC100 and NC200, RC2014 when I get around to building it), but if you start using DOS it shows various advantages very quickly, even on an 8086. Depending on your needs, potentially insufficient advantage for the increased price at the time, but it is a more powerful platform.
I do have a very large soft spot for the PCW, it was my first serious computer, and for the price and the capabilities was one of the best, if not the best, CP/M machine (although granted, Locoscript isn't a CP/M program, and programs such as Micro Design 2 are not exactly very portable, running only on the PCW and CPC)
Before the PCW was around, there was a thriving business in IBM-XT Clones, and my company ponied up £3k for a twin disk model in 1985 for my new PCB design system. I ended up using Jim Button's Brown Bag Software who had a version of PC-Write which was much simpler to use than Wordstar (which I used at home on my CP/M machine). There was also a spreadsheet (PC-Calc ?) which worked as well as the spreadsheet on the Apple //c machines we also had.
The whole shareware scene in the 1980's was amazing, and to my credit I bought proper licences as I was using them for commercial purposes. Yes, I even bought PK-ZIp back in the day ...
Ditto. Wordstar on a Amstrad CPC6128 did my degree dissertation.
This meant that when I was nearly late for the submission deadline, there were no photocopy places open. So I just printed the entire thing out 3 times on my Canon PW1080a dot matrix, in NLQ mode. I did get the 3 copies onto the counter at the bindery with 30 minutes to spare!
Wordstar on a Amstrad CPC6128 did my degree dissertation
A friend did all his degree and postgrad work on a PCW8256 with dot matrix.
Another friend had a ?286? at home but no printer, so brought me raw PCL5 files I could stream to my Canon LBP4. Very posh.
My sister did all her degree work on a Canon Starwriter.
I did a PGCE on a Psion 5mx and an old HP Inkjet. Saved a lot of space in my room in halls and immune to power cuts.
I still write emails in plain text with a monospaced display font where possible :-)
M.
"Ditto. Wordstar on a Amstrad CPC6128 did my degree dissertation."
Pampered kids!
Typewriter, editing with scissors and stapler.
And back in the day, publishing involved galley proofs and page proofs where edits had to preserve the line length to avoid having to reset entire pages. However it did save the day when I was given somebody's page proofs to read and realised there were errors converting imperial to metric (or possibly the other way about) which really mattered because it dealt with sea-level changes. They had been missed in his thesis by both his supervisor and the external examiner, by the journal editor and reviewer and in the galleys.
We had a couple of Superbrains tucked away in room adjacent to the design lab, and they both ran Wordstar for us engineers to write up reports. Being young, ernest and skint, I bought a "Matmos PC" from the back pages of a computer magazine for about £200. Twin floppy discs, one Z80 and a very odd keyboard all running CP/M. Originally made by the German typewriter company, Triumph Adler, they were sold off cheap as they were so slow.
However, I managed to PIP Wordstar from a Superbrain via the serial port to my Matmos, and that was officially my first dodgy copy of any software. Slow as molasses, but at least it worked well enough until I could afford something better.
Mention of TA and weird keyboard reminded me of the (pretty much identical*) TA Alphatronic machines we had at university in the late 80s - they'd been repurposed as terminals for the campus network.
* https://www.computinghistory.org.uk/sec/5078/Triumph-AdlerAlphatronic-Matmos-PC/
I didn't remember dBase-II swapping like that, but now that you mention it... I still have a dBase-II directory here on my Linux box, and I see a file called "dbaseovr.com" - that might be the overlay.
I used it to write 3 different store and/or inventory systems for local small companies. And looking, I see that I still have the source for all 3. Me, a pack rat? :-) Two have files like "012085$i.dbf" - I'm guessing those are test database files.
I also have 3 commercial games: Adventure, Enchanter, OrbQuest, and my own "Explore" game system - way better!!!
I used "VEdit" for editting. But, a friend bought a kit (which I helped build) for a CP/M computer. She didn't have an editor, so I wrote my Ded (Draco EDitor) for her. Not sure she ever used it much, but I switched to using it - I had seen microemacs able to handle multiple files in memory, so I made Ded do that, and that was enough to make me switch. It "compressed" files you editted - turned runs of spaces (like in source code) into a count.
[Author here]
> Aside: if I can do word processing with Wordstar in a 64K memory machine, why am I using hundreds of megabytes for LibreOffice in 2024??
Just for what it's worth, before this was published, I have an article coming on WordStar, getting it today, and how to run it on a modern OS. With a screenshot of me writing the copy, in real live WordStar, on macOS 12...
In 1983, RAM was about $2000 / MB; thus $125 / 64KB.
In 2024, RAM is about $0.00128 / MB; thus $0.00008 / 64KB
So, if your modern electronic typewriter is using:
100MB the cost is $0.12
200MB the cost is $0.24
....
1000MB the cost is $2.40
...
98 000MB the cost is $125
Whereas, in 1983, if your electronic typewriter is using
64KB the cost is $125
So, your 2024 electronic typewriter is about 60 times more efficient than your 1983 model.
> So dBase had two blocks of memory: a core block in memory, and an overlay block
Overlays is how many large programs worked, also on MS-DOS. The Microsoft linker had support for specifying which object files were part of the fixed block, which could be swapped in when functions in them were called. I never did CP/M development, but I suspect it had tool kits that did the same. After 32-bit 386 and its successors became common, large MS-DOS programs started using DOS extenders instead of overlays. Much easier for the programmer, and more efficient.
The Osborne and dBaseII were the start of a business for me. Program and Constants file on Drive A, data on Drive B. Shortly afterwards we made it sort of work on MP/M (which has been mentioned in these responses). A bit of a dog's breakfast especially with the hardware then available. The system was for a newspaper group c1985 and was quickly re-written in RM-Cobol which on NCR's CP/Net Modus system provided proper record locking and the business took off. CP/M had Level 0 to 9 (?). Anything in Level 0 was available to all other levels but the other levels were private to eachother so we could have different routines, utilities, software in different levels.
I've still got the Osborne though it's not been fired up for year or ten.
According to https://eylenburg.github.io/os_familytree.htm there used to be a CP/M descendant called "4690 OS" that was still being supported until recently. It's probably this one: https://en.wikipedia.org/wiki/4690_Operating_System
So I guess nowadays only FreeDOS remains as a spiritual successor to CP/M.
I've mentioned before that my first job was tending a network of CompuPro MP/M 8-16 boxen, which was CP/M with a very small hack that could take advantage of CompuPro's 8088/8085 dual CPU board, and a very large hack that did multitasking.
I don't remember the details of why, but I remember feeling DRI was kind of a bag of dicks.
I had one of those boards in my system. It was a *very* expensive system - 20 slot S-100 box, dual 8" floppy box, Epsom MX-80 printer and a small terminal (Adds Viewpoint?), 1200 baud modem. I didn't have MP/M, but I remember trying it out at a local computer store.
I don't really know why, but I wrote an 8080 emulator for the 8088 side, and it actually ran some of the CP/M programs I had. A couple of games didn't work - maybe they used undocumented BIOS stuff. My Draco stuff *did* work, so maybe I was thinking about make an 8086 version of it - never did, though.
We had an S-100 box using cards from SD Systems. The OS was S-DOS. I'm not sure whether it was a CP/M clone, a ripped-off version or a licenced version re-branded but whatever ti was it did everything CP/M did, ran everything CP/M ran and was functionally indistinguishable. We also had a Microsoft FORTRAN compiler for it.
Nostalgia isn't what it used to be.
“Microsoft licensed it from SCP, renamed it MS-DOS, and licensed it on to IBM, while retaining the rights to sell it on to other companies. The rest is history.”
__21: Unknown error Dave:__
SCP might have assumed they had licensed QDOS. But the terms of the agreement allowed MS to wriggle out of that. Effectively, Microsoft outright bought QDOS (MS-DOS) from SCP and licensed it in a non-exclusive contract to IBM. IBM unknowingly provided the finance up-front for the purchase. Curiously enough ClippyAI has this to say on the subject: "IBM agreed to fund Microsoft’s purchase of QDOS"
ref:
“This license agreement is made and entered into this 6th day of January, 1981, by and between SEATTLE COMPUTER PRODUCTS INC, a Washington Corporation (hereinafter referred to as SCP), and MICROSOFT, a Washington general partnership (hereinafter referred to as MS)”
My memory is telling me that Microsoft started by licensing it and then, once they realised the legal implications might stop them making a shit load of money, bought it outright.
They ended up paying more later, didn't they? Something around not bothering to mention things like "we're gonna resell it to IBM"?
I seem to recall that M$ paid SCP due to SCP taking advantage of having the rights to MS-DOS for their own CPU's. SCP had started selling an NEC chip to the replace the 8088 in IBM PC's and then stating it was an SCP supplied CPU. M$ sued SCP to stop this practice and apparently ended up with a lot of egg on their face.
> My memory is telling me that Microsoft started by licensing it and then, once they realised the legal implications might stop them making a shit load of money, bought it outright.
I _think_ the chronology went:
1. MS pays SCP to design the Microsoft Softcard, for running CP/M on an Apple ][. Developer: Tim Paterson
2. SCP designs an S100 bus 8086 card. CP/M-86 was already announced but not finished or shipping. The board runs MS Standalone Disk BASIC.
3. Some ~6 months later, still no CP/M-86 yet, Paterson writes QDOS, re-implementing CP/M-80 on 8086 but using the MS BASIC disk format
4. MS pays $25K to licence what is now called 86-DOS.
5. ~6mth later, MS pays another $50K for full rights, but SCP keeps an unlimited license to sell it on its own boards, and does so. Eventually it sells "Seattle DOS" with just a CPU chip in the box.
[IBM launches the PC]
6. ~6mth after the PC, DR launces CP/M-86. IBM charges $40 for DOS or $240 for CP/M-86.
[Time passes. Thorin sits down and starts singing about gold.]
6. Paterson leaves SCP and ends up at MS.
7. Paterson writes MSX-DOS for the MSX micros: CP/M compatible but with the DOS disk format and commands.
8. MS buys 86-DOS outright for another $925K.
I would modify "3." to read "re-implementing the CP/M-80 API on 8086".
Since Intel designed the 8086 to be largely source code compatible with the 8080, it was in SCP's interest to make porting CP/M-80 code to 86-DOS as easy as possible. Hence 86-DOS having an utility for translating Z-80 assembler to SCP's assembler syntax and another utility for reading CP/M formatted floppies. The names of some file operations were chosen to be easier to understand, e.g. COPY and ERASE (which was the original 86-DOS command for erasing files). One other difference between CP/M and 86-DOS was the latter recording file size as actual number of bytes.
The general point of the article is correct with respect to why CP/M was so important. Having a standard API for software greatly increased the market for both commercial and "free" software, which then led to a wider variety of software.
I do remember picking up a Digital Research flyer at the First West Coast Computer Faire - along with an original Apple II flyer.
> "re-implementing the CP/M-80 API on 8086".
Yeah, OK, I will happily concede that. :-)
I still don't think it's a _clone_ of CP/M. It's a compatible re-implementation for a different CPU, partly because it took DR _three years_ to ship CP/M-86.
I have long preferred DR's products to MS's, but even so, ISTM the company got complacent, sat there collecting the cheques and not doing a lot while the market shifted underneath and around it. The IBM PC wasn't the first 8086 machine. (Heck, IBM's own DisplayWriter used it over a year earlier.)
It then, belatedly, got busy and had some modest successes, but not as great as it deserved.
If you read the "Oral History of Brian L. Halla" I link to you will find another surprising fact about Dr Kildall...
pip had the source/destination the other way around from copy.
I had a stupid idea once to write a CP/M command interpreter for mobile. So, to send a text message, compose it using ED then enter:
pip <phone number>=message.txt
to send the message. To do a voice call:
pip <phone number>=PHONE:
IMs:
pip <phone number>=CON:
Some of it, but you'll quickly run into the use of PIP vs COPY, the lack of almost any built in commands, absence of directories, and time stamps until CP/M 3.0. I think if you'd started with DOS 6/7 and never experienced early DOS releases you'd struggle.
An early programmer of DOS would note a number of API similarities.
There was also no standard in screen addressing, and GSX wasn't exactly widely used for graphics. For DOS at least text mode had a defined API, and there wasn't a need to enter terminal types or impenetrable control codes to patch programs to do cursor addressing.
"For DOS at least text mode had a defined API, and there wasn't a need to enter terminal types or impenetrable control codes to patch programs to do cursor addressing."
Yes, well, for the time period, there was only MDA and CGA. It's a lot easier to target a single system architecture than to target multiple different type of hardware from multiple vendors in a time before compatibility was seen to be a potentially good thing.
Of course, there were other systems that ran MS-DOS and were NOT IBM compatible, so addressing the screen and/or displaying something required different methods, a different version of ANSI.SYS. And like CP.M, a custom IO.SYS and maybe MSDOS.SYS (equivielnt to BIOS and BDOS) to suit the underlying hardware. My own experience was the Apricot F1, which didn't even have a text mode as such, only a graphics mode. No playing in BASIC and POKEing a value into the C800:nnnn page to put characters on the screen.
86-DOS was running on a variety of machines BEFORE the IBM PC was officially announced in 1981, and there was no API for cursor addressing in MS-DOS, my recollection was that API was an IBM ROM BIOS call using one of the software interrupts below 20H that Intel explicitly warned against using and thus made a mess with the 80286 transition. Conversely, the IBM PC implementation did not correctly implement the MS-DOS get time call, where on an SCP machine the time was reported to 0.01 second.
I guess that not much, but its legacy is still very evident in Windows. Where *nix used - for command line switches, CP/M used /. So when Microsoft added directories and nabbed *nix commands it used \ as the folder separator.
And of course MS-DOS got its 8.3 file name format from CP/M. When I had some Linux files with a character that Windows did not like, it did not just substitute that, but fell back to the Windows95 8.5 name substitution.
[Author here]
> its legacy is still very evident in Windows.
True.
> Where *nix used - for command line switches
True.
> CP/M used /.
*Not* true. CP/M did not have command line switches as standard.
That was the point of my link in the story, where it says:
«
(This influence from DEC OSes is still visible in Windows today.)
»
Ah, thank-you for verifying that CP/M didn't use command-line switches. That raised a bell when I read it. There was nothing stopping an individual program from using them, I expect. I can't recall how a program received command-line arguments, and I'm not going to go look at old Draco sources! I'm busy with more important stuff - watch that web directory in a few days.
I was very very fortunate that the first ever home computer I got to use when I was a lad was a DEC Rainbow 100 which came with CP/M, but not just any old CP/M but CP/M-86/80, needed because the Rainbow had 2 discrete processors, a Z80 and an 8088, so it could run the 8-bit applications written for the Z80 and the 16-bit applications written for the 8088, with the OS seamlessly switching which processor executed the running application. The rationale for this architecture was that there wasn't a lot of software for the Intel 8088 but there was loads for the Z80, so you'd have a lot of choice of applications straightaway. It was a typical DEC product: superbly built, innovative technology but sadly not PC-compatible. There was nothing like it at the time and whilst all my friends had Sinclair Spectrums, Amstrad CPCs and Commodore 64s, the Rainbow 100 couldn't play games and so wasn't much fun for a teenager until I got my hands on some programming books and over time learned BASIC, then Pascal, dabbled in 8088 assembler and then C.
CPM as on OS was very basic, didn't support directories as such but there were up to 16 "user areas" which were essentially "partitions" that segregated files. The default user area was zero and others ranged from 1 to 15, you'd switch between them using a CP/M command (that I can't recall or be bothered to look up) and the DIR command would only show files in the current user area - files only, no support for folders and IIRC, there was no way of copying or moving files between user areas. As soon as MS-DOS became available with it's modern folder support, CP/M hardly got a look in from me, still though, it was a fairly easy intro to OSes and I learned a lot from it so I'm very thankful.
In my professional career, I even got to use PL/M (it was PL/M-86) to write and support firmware for the ticket machines and automatic gates that were introduced to the London Undeerground in the late 1980s so again, my thanks to Gary Kildall.
That Rainbow sounds a fair bit of a beast, I definitely need to look into it!
I can understand not wanting to bother with old knowledge, but work is over now, so let's have a quick decompress looking at old computers. What's needed for this is Digital Research's 'CP/M Plus User Guide' (Google reveals various PDFs).
Changing user area you can either type in 'user 3' , 'user' then 3 when it prompts, or just 3a: (assuming drive A: ). Once in a user area it applies across all drives, so you can't be in 3a: and 2m: simultaneously.
For PIP the option is G
i.e. PIP M:[G0]=A:PIP.COM[G3]
You cannot, however refer to a file in one user area from another. i.e. sit in an empty user 3 area and do '0A:PIP M:=A:PIP.COM[G0]' . What a huge pain!
Built in DIR can't list user areas other than the one you're currently in, the separate DIR.COM can use either [user=all] or [user=<n>]
The non DR documentations is poor - various documentation claims that PIP's G option only works on a destination (untrue), and the PCW manual doesn't list the G option at all (but it works)! It only lists [user=all] for DIR.COM, but I guessed user=<n> would work before confirming it in the official DR guide.
It's entirely possible CP/M 2.2 is even more limited. A quick look seems to show G in PIP still exists, but I don't have a Digital Research code based CP/M 2.2 system sitting around (apart from the PCW, the remainder I think are cpmish or more modern hacked alternatives).
There was the ZPM replacement for the CCP and BDOS, but they tend to concentrate on scripting and development (plus built in character positioning which I can understand), rather than user focused quality of life improvements. CP/M is very limited on memory space, and most systems didn't have the minimum 256KB of memory found in the PCW.
"That Rainbow sounds a fair bit of a beast, I definitely need to look into it!"
LSI did a similar beast. The one I used also had a 4 port RS232C board in it through which we attached 4 dumb terminals for 5 users. The main console also had, IIRC, 4 virtual consoles accessed via CTRL 1-4 (or something, I forget now) but as per the Rainbow, seamlessly ran 8-bit (.com) and 16-bit (.cmd) programs in multi-user. ISTR there was a CP/M-86 for the PC and if you swapped out the 8088/8086 for a NEC V20/V30, it could also run 8- and 16-bit CP/M programs because the NEC V20/30 chips included the Z80 instruction set. (Very vague memories of this, corrections welcomed!)
Funny how the various executable file extensions were used in a variety different OSes, though the following mightn't be 100% correct, but
CP/M - 8 bit executable was a .COM file, 16 bit was a .CMD file
MS-DOS - an executable that limited itself to a 64KB address space was a .COM file, otherwise it was a .EXE file
VAX/VMS - a .COM file was a Digital Command Language (DCL) script, .EXE was an executable
Windows - .EXE is an executable, .CMD is a CMD.EXE script (CMD.exe also recognises the .BAT extension for batch files)
The early versions of Turbo Pascal for MS-DOS generated .COM files, later versions .EXE files and IIRC Turbo Pascal for CP/M only generated 8 bit .COM executables
I ran a small department in 83/84 that got CP/M running on my own Z80 hardware. It was a fantastic buzz when we got WordStar up (not to mention C and Fortran compilers). It really did feel like the world had changed at the time. IBM eventually screwed us, of course, because people still remembered who they were back then.
I've done a huge amount of hardware since then, from bit-slice to ASIC to quantum, but nothing really compared with that.
Thanks for the name check! Interested parties might be interested to know that a year or so so I did a complete rewrite of CP/M for the 6502 --- it now runs on most mainstream 6502 micros including the BBC Micro and the Commodore Pet. It works really nicely, and includes relocatable binaries, a (rather bad) assembler, a port of Atari Basic, a version of a 1970-era Pascal compiler/interpreter, and a few other things. I ended up making a few changes to the CP/M API but the core concepts translated very well. Now I'm wondering about an 8051 port...
CP/M Gary Kildall
dBase Wayne Ratcliff
BDS-C Leor Zolman
C Dennis Richie/Brian Kernigan/Ken Thompson
MS-DOS Tim Paterson (apologies to others!)
yacc Stephen C. Johnson
.....and now we get Elon Musk asking for fifty BILLION dollars in personal compensation!!!!
In the days when CP/M was invented and released, everyone was just working to help users start to use and understand the computer systems, originally making money was only seen as a bonus, not a requirement. I had a friend who worked to setup an S-100 system in a clinical analysis laboratory, only trying (and succeeding) to help the physicians document how children were walking with problems so that they could be surgically fixed.
Once he had his plans working, he gave the details to other organizations working to help get the kids walking better. All of today's computing is a mainly a result of those early days years ago with everyone just trying to get everything being fully usable and working well for everyone else - I am so impressed and happy with everything I was taught back then by virtually everyone, but so many of my associated people in that world are deceased now - technicians, physicians, physical therapists and surgeons.
When I started getting in to building hobbyist computers a few years back, the CP/M version in ROMWBW was the only real way to go, so I have several such systems sitting around. I don't have much use for them (the building was the fun part, fighting with CF card formatting was not) and there seem to be far more options now, but I still keep a little Easy_Z80 plugged in via USB-Serial just to poke at it now and then.
> cobbled-together demo OS created to help sell some model of hard drive.
*Floppy* drive, maybe. It predated hard drives on personal computers.
It could be true. Maybe he, or the company, didn't want to own up to that?
But while CP/M-80 _was_ extremely small and minimal, I think we need to bear in mind 2 things:
1. It had to be. The early CP/M machines only had a few 10s of kB of memory, because it was so expensive, and one or two floppy drives holding ~100-200kB.
2. This was perfectly normal for OSes in the early 1970s. The DEC OSes that inspired much of CP/M also ran in single- to low double digits of RAM from media barely holding triple digits of kB, so they too were tiny simple things, a few kB in size.
DEC OS-8 for the PDP-8 took up 256 words of RAM.
DEC RT-11 on the PDP-11 ran off 2 x 250kB disk drives, and in 56kB of RAM supported 8 concurrent users.
RSTS/E on the same machine supported 16 users. Max program size, 16kB.
By the standards of mid-1980s DOS PCs, they were tiny and toylike, but in 1970 they were production-grade pro systems.
RSTS/E,
"By the standards of mid-1980s DOS PCs, they were tiny and toylike, but in 1970 they were production-grade pro systems."
In this entire discussion, that it probably the best takeaway point, especially for our younger readers. Even at the time that CP/M was a mature OS, home computers still generally came with 8 or 16KB of RAM at best, with the option to upgrade to 32K or even <gasp> 48K! They normally had a ROM BASIC plus i/o & video RAM taking up chunks of the memory map too, so getting 64KB to run CP/M often wasn't even an option in those early days of microcomputers when comparability was rare and everyone was inventing their own way of doing stuff because each one was a "first" in the field (or at least a variation.) There was no "proper" or "expected" way of doing stuff. It was only when the microcomputer market was relatively mature and ways of doing bank switching (and RAM pricing) became economically viable to get 64KB of RAM and CP/M running on Z80 based micros other than hugely expensive business machines. Fun days when a single individual person could actually understand all of the underlying hardware and the code the ran on it.
"Even at the time that CP/M was a mature OS"
Yes, CP/M was a mature, serious OS, not just a toy. The whole microcomputer revolution was a marvel for those of us who had a use for which, up to then, a mini would have been the answer but no budget for one. People like myself, working in a lab and with previous experience with FORTRAN and punched cards on mainframes. In fact it was likely that not even a mini would have done the job that could be achieved with an S-100 box. There were all sorts of cards available such as ADCs to aid interfacing with instruments. I build a microspectrophotometer and, realising that the 9-bit ADC wasn't enough, added another 4 bits with an op-amp and a 4-bit CMOS switch and a stepping motor controlling the continuous interference filter. Microsoft FORTRAN had I/O equivalents to POKE and PEEK (PUT and GET IIRC). I'm not sure one could have added the necessary interface cards to a mini.
Some of my colleagues went on a computing course which taught Pascal so I changed to UCSD Pascal which ran on the same H/W but subsequently on an IBM PC clone. Maybe another subject for an article Liam?
Hmm. You sure about CP/M on the COCO? COCO generally referred to the Radio Shack Colour Computer, which had a Motorola 6809 processor (it had an 8x8=>16 multiply!). I don't recall CP/M being ported to anything other than 8080/8085/Z-80. I had a COCO for a while, and it essentially just ran BASIC and games. The 6809 was a couple years after the 8080.
My first real programming job was on CP/M. Having spent a few years at uni on DEC machines I created a multi-user game on the uni's two CP/M micros using a shared floppy drive. Coordination was by locking a file as if it were a semaphore and the code would then act as leader or follower. Fun times, but on the strength of that landed a part time job creating a stock and invoicing system. The real issue was a non-technical sales manager that piled on creeping featurism so that every version we sold was different ... cue maintenance nightmare and me leaving to find a job in mainframes. Early micros were cantankerous physical machines but much less sanctimonious than the whole priesthood thing that was growing around the larger boxes. MP/M was floating around, but no-one could really see a use for it.
[Author here]
> The Amstrad 6128 came with both Dos 3.0 and CP/M: on floppy disks. Happy days.
I think you are confusing different machines.
The CPC range did not have a DOS as such, and definitely could not run any form of MS-DOS or anything like it. (Although the BASIC ROM could access disk drives in a CP/M compatible fashion.)
They did optionally support CP/M, though, and so did Amstrad's ZX Spectrum +3.
I never owned either but I think it was CP/M Plus, which is CP/M 3.0, the final version, which supported file timestamps and a RAMdisk made from banked memory.
Perhaps you were conflating the CPC with the much later Amstrad PC1512 and PC1640, which were 8086-based PC-compatibles. They came with both DR DOS-Plus and MS-DOS.
RAM disk from banked memory? I guess its inevitable that someone else did that. I was quite proud of my version (no, it was never released - it depended on the DMA of the specific floppy controller, and the bank register of the specific dual-CPU card). And, it made using the machine much better - sooo much less disk drive whirring and thunking (mine were big 8" ones)!
I had a CP/M machine back in the dim and distant past. The first was a Newbrain and the second an Apple IIe with a Microsoft Softcard. And CP/M made them compatible - but only sort of. Yes, they could run the same software (although sometimes, as with Wordstar, it was necessary to run a configuration program to set up the screen correctly first) - but, astonishingly, they couldn’t actually use the same disk format. So to copy files and programs from my old Newbrain to my newer Apple I needed to squirt everything over serial. Despite both having 5.25” disk drives.
I was incredibly envious of those lucky so and sos who had an Amstrad PCW. Yes, Amstrad, purveyor of all kinds of tosh and nonsense also made the best CP/M machine of them all - and given that the disk formats weren’t compatible anyway it didn’t matter that it had an odd 3” disk drive.
"Amstrad ... made the best CP/M machine of them all"
That would depend on what you wanted to do with it but I don't think you could add the sort of interface boards that some of us needed. I suppose the difference is between a machine built around CP/M as opposed to a computing machine which could run CP/M as its OS.
The PCW had a pretty wide selection of interface boards, but its major limitation is that it only had one expansion slot which expansions needed to be piggy backed on to. Expansions included :
serial/parallel ports (but typically only *one* port)
extra memory
faster CPU
light pen
mouse (at least three types)
hard disk
networking (bit rare and non standard though)
graphics tablet
scanners
soundcard (generic YM) and joystick port
I *think* it may have had some ADC converters too
There are expansions that provide multiple slots rather than piggy backing, but that's all hobbyist.
Buiding on top of CP/M, Locomotive Software wrote a wordprocessor which could do around 92 characters per line, more than the standard 80 for cheap monitors and enough for a full line of text - halfway to WYSIWYG - by playing games with the characteristics of the cathode ray tube's hardware. It could do the same for a dot-matrix printer. Another feature was the large number of dedicated function keys, a bit like a PC keyboard with half a Speccy grafted alongside. So it incorporated its own specialist hardware drivers and had to run a customised CP/M on equally custom hardware.
They called it LocoScript, sold it to Alan Sugar and the Amstrad PCW became the first mass-market office wordprocessor, cheap and easy enough for every SOHO klutz. All the other hardware makers wanted it - but their universal 80-column 640x480 screens castrated it. Without portability, this particular embedded CP/M-like OS struggled on in the PC market for a while, but its lack of hardware portability - the very reason CP/M was created and so much copied - was its undoing. You could boot real CP/M on the PCW and run the usual stuff too, but that was a level of expertise above the average office typist and competed in the main market, where better machines - and MS/DOS PC's - could be had. Even wild success could not overcome its USP's lack of portability.
[Author here]
You are misremembering badly and getting a lot of details confused.
The Reg has done a great and encyclopaedic history of the CPC range:
https://www.theregister.com/Print/2014/02/12/archaeologic_amstrad_cpc_464/
And of LocoScript and the PCW:
https://www.theregister.com/Print/2015/09/09/joyce_turns_30/
So, to pick a few nits:
> Buiding on top of CP/M, Locomotive Software wrote a wordprocessor
Nope. It didn't run on CP/M or any OS. Locoscript booted on bare metal.
> could do around 92 characters per line
90 × 32 text, 720 × 256 pixels.
https://en.wikipedia.org/wiki/Amstrad_PCW#Technical_design
PCW hardware standard. CP/M and its apps ran in the same res.
Compare with the Hercules Graphics Card mono res: 720 × 348
> this particular embedded CP/M-like OS struggled on in the PC market
(?) CP/M didn't struggle in the PC market, and LocoScript also ran on top of MS-DOS.
https://archive.org/details/locoscript-pc-v-1.06
But while it was excellent, it appeared just when Windows 3.x was killing the market for DOS apps.
Thanks Liam, I see many thoughts resulting from my "misremembering badly and getting a lot of details confused" these days too, any every time they make me rethink what I remember doing originally. I suspect this environment is just a result of all us early workers recording very little documentation originally and having to work in so many different operating system environments as they all started to appear ... I was using CPM at home, RSX11M at work for our customer support, while the company accountants needing their Apple and MS-DOS problems fixed most days too. - but I was initially only employed for electronics device development so it was just new things for me to do.
This was originally a fairly common messy working world everywhere for everyone as operating systems started to appear (the icon is my memories).
>[Author here]
You are misremembering badly
Not this time. I was there.
> Nope. It didn't run on CP/M or any OS. Locoscript booted on bare metal.
Oh yes it did. It incorporated a modified form of CP/M in its codebase. That was how it could do things like manage processes and access disc drives. Richard didn't code /everything/ from scratch.
(?) CP/M didn't struggle in the PC market...
Well, it didn't exactly outshine MS-DOS takeup. But my point is that the customised CP/M underbelly of Locoscript fared even worse.
...and LocoScript also ran on top of MS-DOS
Locoscript PC had to be castrated down to 80-column display to fit VGA, ruining its full-page-width semi-WYSIWYG qualities. And of course the standard PC keyboard didn't have enough function keys to remap everything cleanly. It was for these reasons that Locomotive took so long to decide to do it anyway. Had PC hardware been up to a full-featured Locoscript, history would have been very different.
You should be interviewing me not dissing me.
But thanks for the interest, anyway.
The fact Locoscript has modified CP/M in its codebase is news to me, but makes sense and answers some questions I had about why Amstrad spent effort porting CP/M in the first place (although it did make a fair bit of sense).
I have to question other assertions though. Given the PCW only has four function keys (F1-F8 using shift) it is clearly possible to remap AT keyboard keys, albeit possibly slightly awkwardly, but that's always a limitation of custom hardware (or at least a custom keyboard).
I do wonder how well an 8086 Hercules or VGA PC would run Locoscript PC as the PCW benefited from its implementation of roller RAM.
Leaving that aside, whilst Locoscript was a decent package it was not revolutionary (sorry!) and at least for version 1.x - which is what I wrote a lot of college work on, not stunningly fast either. It was easy to use and relatively feature complete.
The whole PCW package with bundled Locoscript and CP/M was a compelling proposition at its price point. Lots of decent, well priced software on a system architected to render text quickly (and with roller RAM, bold and italic are possible), lots of memory for an 8 bit, an OS with plenty of historic software available, high resolution graphics, and expandability.
It's arguable if the 8086 PC offered sufficient advantages over the PCW given the price disparity and functionality, it really depended what software was run. By the time the 286 arrived it was substantially more clear cut in favour of the PC.
I don't think history would be substantially different if Locoscript had been released earlier on the PC (unless Amstrad could produce a similarly priced cheap PC/printer/software bundle), but Amstrad, Locomotive, and Creative Technology have a lot to be proud of.
The PCW also had about 10 dedicated function keys for all kinds of common wordprocessor actions. Like the Speccy these were named, so mappings to the PC keyboard's F9-F12 and [Ctrl]-x were not immediately obvious.
Locoscript and the PCW hardware were tightly integrated. So things like display output had to be completely rewritten. By the time Locoscript PC appeared, the PC had already moved on a generation or two, so things were not quite so dire.
Don't forget the integrated printer. Same proposition; direct drive, no standard interface chippery, cheap and easy enough for any typist to use. A lot of offices never booted true CP/M, just used the wordprocessor bundle as the office typewriter.
Locoscript PC was never going to fly until SVGA introduced 800 x 600 display. But by then it was far too late.
So......my CP/M-80 machine had a tool called "MediaMaster"................................
MediaMaster could write floppies in a gazzilion different formats....KayPro, MS-DOS.........
So....I would write dBase programs on my Osborne 01....and test them elsewhere on an IBM PC!!!!
I still never even booted up the Z80 processor onboard, although I'm told that every power-on causes the Z80 to check something & then transfer control to the 8502. I been running it since 1985!
(I saw on youtube somebody logging their C128 into CP/M and even with a hardware-emulated disk, it's ponderously slow. Plus, what could you run on it? Furthermore, you know what, I never had the 80-col capable monitor anyway!)
Nobody can deny Kildall's triumphs, contributions, and unwarranted downfall.
Nevertheless today I am far more impressed by KolibriOS: really amazing stuff!
The fascinating part of the CP/M systems was the wide variety of hardware. Even for a serial port (let alone the floppy or hard disk controller, if you had one) you could have a polled system, one with interrupts, all the way up to full DMA (yes even serial ports with DMA on a few models). Since you had a BIOS the vendors were free to make as "cost optimized" or high performance of hardware as they wished to.
And since the BIOS was nicely seperated they even had network filesystems -- no ethernet yet so these tended to run on a multi-drop serial setup. Given the systems had like 64KB RAM anyway the relatively slow performance of filesystem at serial speed was not as big a deal as it sounds like it would be.
I'll be honest, i used one once for a few minutes as a child, and have a friends kaypro in storage (but no software) so I have no first hand experience, but nevertheless find the variety of hardware they had refreshing.
I was very lucky to have the opportunity to develop some comms software on a CP/M 8 bit Superbrain to capture the output from a PDP 11 (running a traffic control system) and save the output to floppy for later analysis. As my knowledge of CP/M grew I discovered there were a few bytes of the memory map unused. So I pranked a colleague's machine by inserting some interrupt driven code that inverted the colours on his (monochrome) screen once every minute. His reaction? Just carry on as though nothing had happened!
I suspect that would be frowned upon nowadays...
I’m rather late to the party, but I only just came across this article. And it’s as good an occasion as any on which to get straightened out by someone regarding something I’ve been unclear on for decades.
I understand, broadly, that CP/M was a big success in the early days of 8-bit personal computing. The article explains this nicely:
‘CP/M . . . . became one of the first cross-platform microcomputer OSes, and revolutionized the hardware and software industries.
. . . .
Over the following few years, multiple floppy interfaces were built and sold for early 8080 and Z80 machines. CP/M was the de facto standard OS, partly thanks to how Kildall adapted the OS [by concentrating the hardware-dependent portions of CP/M in "a separated Basic I/O System (BIOS)".]
. . . .
According to Kildall: "CP/M was an instant success. By 1980, DRI had sold millions of copies of CP/M to manufacturers and end-users." It ran on over 200 different machines, and cost $70 a copy. By 1982, DRI's revenues were $20 million a year.’
(Emphasis added.)
So the strength of CP/M was its portability, and its resulting ability to run on most 8080- or Z80-based machines.
(With or without an S-100 bus.)
But what machines are we talking about here?
Most of the biggest-selling 8-bit personal computers that I know of were not (primarily) CP/M machines, and I don’t know what the CP/M-using machines are that apparently ultimately outsold them.
(At least in the aggregate — even if no one of them, by itself, was among the top sellers.)
* Most of the big-selling early 8-bit personal computers that I know of — such as the Commodore PET and its various successors, the Apple II, and the BBC Micro — used the 6502 or one of its later variants or descendants, and therefore required an add-in card with an 8080 or a Z80 — typically a Z80 — if they were to run CP/M and CP/M apps.
* And at least in the earliest years, even most of the biggest-selling and most prominent personal computers based on the Z80 did not use CP/M — at least not as their standard or primary operating system.
(Details regarding the use of CP/M on a TRS-80 Model I or III or a Sinclair ZX80 or ZX81 — let alone on an Apple II or a Commodore 64 — are complicated. And I believe that doing this always required supplemental hardware: on the 6502-based machines, a Z80 add-in card or cartridge, and on the Z80-based machines, an add-in card or whatever that would modify the underlying system’s use of memory in order to accommodate CP/M’s memory use.
Details regarding the use of CP/M on a TRS-80 Model 4 or a Sinclair ZX Spectrum — or even a Commodore 128 — are not nearly as complicated. Although with the C128 there still were issues — chiefly involving speed.)
So where — on what platforms — was CP/M dominant?
Especially by 1980?
(The C64 and the ZX Spectrum were introduced in 1982, and the C128 was introduced in 1985. Even the ZX80 and ZX81 were not introduced until 1980 and ‘81, respectively.
The TRS-80 Model III was not introduced until the middle of 1980, and it had the same memory map-compatibility issues with CP/M as the original 1977 Model I.
The Model 4 was the first TRS-80 model with the ability to run CP/M without modification, and it actually had multiple ways of running multiple different versions — customized or generic — of CP/M. It appears to me that Tandy probably specifically (re-)designed the Model 4 to accommodate CP/M (while retaining compatibility with the Model III). But it was not introduced until 1983.)
In response to my query, Google mentioned the Altair 8800 — but not the IMSAI 8080! — and computers from Osborne, Kaypro, Amstrad, Cromemco, NEC, Xerox, Epson, Zenith and Heathkit, North Star, TeleVideo, and Morrow Designs. I was already at least cursorily familiar with most of these, but could all of them (plus others omitted by Google) — collectively — have been bigger in aggregate unit sales than Tandy, Commodore, Apple, Acorn, and early Sinclair, taken together?
