CP/M and DOS

CP/M, as many people know, was written by Gary Kildall and based on his earlier microcomputer work, GK had done, namely PL/M. CP/M denotes Control Program for Microcomputers, PL/M denotes Programming Language for Microcomputers. The best story I know - about him, CP/M and his company Digital Research - is here.

CP/M86, which everybody thought to become the standard OS for the Intel side of the micropocessor world was not ready to ship plus it's probably true that Gary snubbed IBM when they wanted to negotiate a good deal for their upcoming Intel 8088 personal computer.

The "not-readiness" was what lead Tim Paterson to write a reduced clone of CP/M under the name of QDOS to make a development board with 8086 hardware work.

MS-DOS (PC-DOS) 1.0 felt a lot more like CP/M than what we know under that name now. Interestingly MS-DOS 2 seems to have been a nearly complete rewrite modeled more after Unix (Xenix) the sourcecode and rights for microcomputers of which MS's Paul Allan had licensed in '79 or so and which, running on Motorola based hardware, was the most important OS inside MS through the 80ies. MS-DOS 2 supported hard disks and hierarchical directories. It seems though that for a number of reasons MS didn't go or get very far in the process of cloning Unix.

CP/M however had been modeled by Kildall after a DEC OS for PDPs, probably TOPS-10 or -20. When CP/M 86 was ready to ship it was a lot more expensive than DOS and although it might have been better than DOS in some ways, the whole SW cycle was just starting anew with software like Lotus 123 that was tightly knitted to hardware, the IBM basic input and output system and MSDOS functions calls. Lotus outperformed every other spreadsheet in its day. And spreadsheets, lest we forget, was what carried personal computers to its first wide spread successes.

Still it can be said that in terms of OS paradigm development it was PC-DOS that made the Unix paradigm win out over all other proposals in the world of small computers, namely the DEC TOPS and VMS models. plink, nix,    praise or blame!
 

Working forward through PC history

Hopefully I'll be back to now (2001, the year of the space odyssey) soon. While surfing around a bit othis afternoon, I found a very nice page that holds images of Interface Prototype Sketches for the Apple LISA.

These pieces are all out there, but often not linked from any easily accessible place and man, more than you can ever read ofi. In this way full text search is and will stay one of the most important tools in the web. A pity in many regards, but then again, maybe not.

Also today, in a post on Bill Gates and open source that Robert Cringely did for PBS and Hannes linked to today, Paul Allen gets mentioned. I'll take the freedom to quote that piece:*

"Your (reader) challenged Gates' claim by noting that the (IBM BIOS CHIP) is highly proprietary. He didn't distinguish the copyrighted chip from the interface architecture it implements. The chip is indeed copyrighted and could be infringed. The open architecture it supports was extended by Paul Allen's DOS 2.1 to actually support dynamic addition of features and capabilities at run time. This (DOS +BIOS) open architecture has been public domain since it first shipped ( Byzantine, but open).

As I have been mentioning here and there, it seems get more obvious that at MSFT Paul Allen was the advocate of an "open" attitude while Bill tended to propietary. A similar pattern as with Wozniak and Jobs at Apple. In the beginning of the PC industry the described sort of "open attitude" was crucial. When IBM left it with their microchannel architecture that was the beginning of their long fall. It is a bisless so today. But it seems it holds on insofar as openness to loyalists and on the APIs sid is still required and that this partial and painstakingly controlled openness is where everybody with a business target is moving now.

I'll mention something else from that article. The Apps group at MS wrote and used a C-Compiler that compiled to p-Code and the p-Code VM. They expected the HW/OS world to become more diversified and to really need a virtual machine target. MSFT was a really early adopter of VMs in the PC world- They could compete against Visicalc on those grounds but not against Lotus 123 that was tightly knit to IBM PC-Hardware and a lot faster performing. It also seems that this performance head start that Lotus got from ingenious Assembly Language programming was one of the things that forced everybody else to copy the PC closely and made only superficially compatible clones from manufacturers like DEC and Siemens obsolete in a short while.

And lastely: Bill was not the person who did move the PC world to more open ore more proprietary, I think, he looked at the trends and decided for implementation strategies to meet those trends. I guess this has been the inner engine of MSFT for their first 25 years: to be not the kings of design but the emperors implementation. Is this the correct abstraction? Yeah, I guess so. They are unbalanced already, MSFT, that is.

*Later note: all 3 documents this article links to are not online anymore. So it goes on the web.

EUP , Dave Thomas, heteroclinic orbits.

Got to know a fascinating and very nice guy yesterday evening at netobjects.evening:

Dave Thomas, canadian, CS professor, founder of OTI and Bedarra He may look not very nice or cool on this picture, but he is a nice guy. If you don't believe me from looking at the picture, ask Mariann.

I didn't hear his talk about EUP this morning because I sat in on Mariann's interview (soon to hear in oe1. we should stream this) with Adele Goldberg which was an opportunity I could not let go. She was my hero for many years.

So I do not know anything about EUP but that P means Programming and it's different from though related to XP. Mr. Thomas seemed to have had an influence on that too.

I heard his talk on teaching Computer Science yesterday though. If we only had had such teachers 20 years ago.

And yes. We reinvent and/or reimplement nearly everything these guys did many years ago because nobody taught us how to do it right. This might be good, I'm not sure yet

Two things to research in the next days: EUP (have been into that before) and heteroclinic orbits (don't know practically nothing here), that Brian Foote spits back on us all the time and that I don't actually know far too little about to discuss it with him.

a little later: I think EUP is End User Programming look for an abstract of Dave Thomas' keynote. or it might be Enterprise Unified Process (EUP).

even later I'm definitely not going to say anything on New York here!

The 4^th Bubble, part 1

Virtual Machines and physical machines :: HAS TO BE REWRITTEN.

The 4^th Bubble starts by looking back in personal computing history. Looking back is not a goal on its own but a way of connecting snippets into a meaningful chain. The meaningful chain allows for the right conclusions && for the right actions. So on we go:

As you can easily derive from the history of personal computing Steve Wozniak designed the Apple II as an open personal affordable not networkable computer and published the Apple II Reference Manual that contained specs of the memory layout, described how to add hardware and peripherals etc.

At the time when business need and other influences started to kill the concept at Apple (1980) Don Estridge of IBM started to lead their PC Division and together with a dirty dozen of design and manufacturing engineers copied Steve's concept in most every detail, switched from MOS Technology to INTEL thereby saving the idea of open affordable personal computing, surely something IBM had not envisioned. GOOD can come from the BAD! After his death in '85 IBM went back to their old ways, started PS/2 - OS/2.

In the same way Apple went on to develop the LISA and the MAC, closed down on information and tried to bind SW closely to propietary HW, a concept all Manufacturers except maybe MITS with their DIY product Altair had followed before Woz.

Both of these counterreformations never found any remedy against the somehow more open concept of the so-called "industry standard PC architecture". So gradually but slowly the leadership in the development of micro and other computers drifted over to the two most decided and competent supporters and beneficiary of that architecture, Intel and Microsoft.

In the meantime a group of people at office copier machine dominator XEROX's PaloAltoResearchCenter started to really build a machine in SW the instructions of which didn't look like

LDA CC1A
ADD #00FE
STA CC1C

anymore. They called those constructs a virtual machine. These virtual machines could and should be implemented on different kinds of hardware but in reality they executed on personal computers like the Alto, the Dolphin and the Dorado. Nobody could afford those fancy machines, practically everybody in IT used a TTY then. The Sun-1 workstation appeared in 1982, 5 years after the Apple II and 1 year after the IBM PC. Even a workstation like the sun could barely run and not really run a Smalltalk Virtual Machine. It took many years to have the kind of virtual machine an Alto could drive run on a personally affordable computer.

The PARC people were not the only ones to think about and develop with virtual machines. One's gotta mention the UCSD P-System that you could install on the original PC with the benefit of having Pascal instead of Basic. Behind that compiler was an OS come virtual machine. And what's also gotta be added, is, that from at least the point on when the Bell labs people rewrote Unix in C, the Unix kernel has become a virtual machine, the model of which in the meantime has grown to be the most pervasive of all VMs.

Third Bubble [deprecated], 1st snippet

No writing on Friday, still lost in balls of mud. No writing on sat and sun, if i can i don't touch this hellish machine on WEs. Took me 7 years and a heavy breakdown to learn it. Move your hands and feet, people!

The third bubble is all about losing the empire and really doing so. I was inspired to that by a story about an Italian teaching hardware/software co-design at some Californian college. Sure I'm gonna get back to that.

As you might see the the third bubble is an expression used for many different things. I might as well call it the fourth bubble and be out of dialectics and into tetrades.

For at least 60 years it has been a rare occasion that logical engineers talk to electrical engineers. Intel did this better than anybody else although not well. One of the reasons we had to wait for Linus is that all the stuff before was somehow bound to expensive hardware. Even the Mac is too expensive and too closed and hardware bound. Concentration on software and licensing helped MS win out over their most potential rival for domination in the PC world. Make no mistake, everybody at Apple but Woz wanted to dominate. What can we do to find an edequate WLC/WCN encoding and beat the WTA encoding that is so favoured by interoperability and the current hard- and software economy?

Remember, from the next time on it's gonna be the Fourth Bubble.

PS: koxinga! Gib uns ein Lebenszeichen! PPS: back to the mud now!