Posts by Richard Plinston 2608 publicly visible posts • joined 27 Apr 2009
> 1. The Apple II did not use variable-speed drives.
Yes, they did. https://collections.museumsvictoria.com.au/articles/1087
> hard sectoring was not because controllers were too slow to do soft sectoring.
https://en.wikipedia.org/wiki/Hard_sectoring :
"""Data read and write is faster in this technique than soft sectoring as no operations are to be performed regarding the starting and ending points of tracks."""
One is faster the other is slower!
> Most commercial CP/M applications that were sold for MSX were re-written for compatibility and to take advantage of the much richer feature set. Borland's Turbo compilers come to mind.
Turbo-Pascal 3.x, for example, was produced in four versions: CP/M-80, CP/M-86, MS-DOS and PC-DOS. The first three had a configuration program which set the terminal type used. The MS-DOS version could be set to 'ANSI' and this would work on an IBM-PC or close that had ANSI-SYS configured in CONFIG.SYS. But other terminals could be configured for non-IBM-PC machines (DEC Rainbow) or those using serial terminals (Wyse 60, DEC VT52, etc). The CP/M-xx versions had dozens of terminal types to select from.
MSX-DOS 1.x could run the standard CP/M-80 version without a 'rewrite'. It did need to be on an MSX-DOS disk though. It seems that it was rather limited and could only use console mode (ie scrolling text) because MSX-DOS did not support ANSI.SYS. Certain libraries, such as Popolony2k, were available that used INLINE assembler code to access MSX facilities.
> MSX-DOS was in no way based on CP/M and provided effectively a clone of MS-DOS
What you say is nonsense because MS-DOS itself is 'based on CP/M' design (if not on source code) because it implements the same BIOS/BDOS/CCP structure and it implements the same CP/M API with only minor variations. It even supports the same 'page zero' structure (in the PCB on MS-DOS) including the FCBs and the 'call 05H' interface. MS-DOS 1 could only run .COM '8080 mode' programs which were structurally identical to CP/M .COM programs.
While MSX-DOS was not 100% compatible to CP/M it could not run MS-DOS programs at all, not even .COMs.
> CP/M compatibility was at best an afterthought
No.Wrong. CP/M compatibility was the _whole_ point of MSX-DOS (and of MS-DOS 1.x). Without it there would have been _no_ application programs that would run on it, at least in the first several months, and no point for anyone to buy the system.
> CP/M floppy formats were vendor specific
CP/M came out when 5.1/4 disk drives were very new and the electronics for drive controllers were slow and expensive. This meant that 'tricks' had to be used to get the best performance for the least cost.
Some used variable speed drives (eg Apple II), some used hard sectored disks with 10 index holes (eg North Star) because the controllers could not keep adequate timing from a single index per revolution. Most could not read or write sequential blocks, they were to slow to be ready so if the blocks were sequentially numbered around the track it would take 8 revolutions to read a complete track. By interlacing and numbering as, say, 1,4,7,2,5,8,3,6 it would only take 3 revolutions. All these were vendor specific.
By the time the IBM-PC and MS/PC-DOS came out 7 years later the controller electronics were much more capable and these tricks were no longer required.
> The only improvement MS-DOS made over CP/M was a single disk format.
The IBM PC originally had 5.1/4 160Kb disks, 320Kb double sided, then 180Kb 9 sector and 360Kb double sided. The IBM PC Jnr had 3.5inch 720Kb while the AT had 1.44Kb.
While PC clones followed the IBM formats. other manufacturers made MS-DOS machines that were not clones and they had a variety of different formats. There were utiliiites that could read these various different formats, just as there were for CP/M, granted there were much fewer for MS-DOS.
With hard drives MS kept the 32Mbyte paritition limit well past its 'use by' date and this led to various PC manufacturers (Compaq, Wyse, etc) and drive makers changing their versions amd/or drivers to support larger disks in incompatible ways. It was IBM that eventually rewrote the code into PC-DOS 4 and passed this back to MS to make 4.01* that finally made compatible hard disks with >32Mbyte partitions.
* not to be confused with MS-DOS 4.0 and 4.1 which were [sort of] multi-tasking versions also referred to as 'European DOS' because they were used by Siemens and ICL as well as Wang.
> So, yes, DR-DOS is 100% a derivative of CP/M.
It is actually completely irrelevant what you consider to be 'derivatives' for the purposes of this license. The only relevant usage is what DRDOS Inc thinks is "CP/M and its derivatives' and that semms to be entirely CP/M itself plus, probably, CP/M-68K and CP/M-Z8000.
> [0] "PC" defined as a machine designed to be used by a single person,
The first computer that I 'used' was a small mainframe ICL 1901. I often ran it alone, changing the tapes and disks, loading the paper in the printer, the cards in the reader and starting the programs. You seem to calling this a "PC".
The first computer to be _called_ a 'personal computer' was the Apple II in adverts in 1978.
> No, CP/M wasn't an OS. It was a glorified program loader.
It never claimed to be an 'OS', it was a Control Program, that is what 'CP' stood for. It was, however, a subset of an OS, a DOS: Disk Operating System, providing that functionality without the other parts that would make it a full 'OS'.
> And (IMO) MP/M.
MP/M provided many more facilities including multi-tasking, multi-user, bank switching ,networking (DR-NET), RAM disks, and, no, programs could not 'take control' of all the hardware.
> My last cpm experience was Dr dos,
An interesting comment because DR-DOS could not run CP/M software
> they could network well before dos or windows could.
MP/M was originally developed in 1978 to be a network server for CP/M workstations using DR-NET. As it was also a multi-user system that could be used with relatively cheaper serial terminals it found its niche there.
> And it was Xenix in name only
Microsoft had licenced the use of Unix source code in 1978 but did not have a licence to use the name UNIX so it called it XENIX. Yes, that was its name. MS announced the product in August 1980.
> Everybody (nearly!) used existing source code as a reference and framework
Microsoft was very protective of its source code. While many were implementing some form of BASIC, usually incompatible, they were not doing so by using MS source code. The implementations of MS-BASIC on many different machines was done by MS themselves or under strict licence conditions. Even DRI had its 'Personnal BASIC' which owed nothing to MS source.
You should note that Bill gate's letter was complaining about the distribution of the binary loadable BASIC interpreter (initially on paper tape) and not of the source code which was only ever on DEC machines where it was cross-compiled to the target machine.
> Just as MS did later with AT&T Version 7 UNIX, which MS distributed as Xenix ...
Actually, MS Xenix predated 86-DOS/MS-DOS by a couple of years and wasn't 'later'.
> and everybody+dog did porting BASIC to all and sundry.
'Porting' usually refers to using a particular set of source code and modifying it to run it on a different OS or machine. The BASIC language was originally developed by Kemeny and Kurtz in 1964. Many somewhat different languages were later developed and called BASIC even if they were fundamentally differernt from K&K's. While MS did *port* their MS-BASIC to many different machines, including the Apple II (in conjunction with Applesoft) i doubt that company could be called 'everybody+dog".
The word you were grasping for is probably 'implementing'.
> The FAT's borrowed from M$ Disk BASIC were very different from CP/M's file handling.
Yes it was and it was very slow on large random accessed files. I developed COBOL applications for business that could run on DRX (DRS20), MP/M, Concuurent-xxx. These used ISAM files for their data. When ported to MS-DOS they ran really slow compared to the DRI based systems. The problem was traced to the FAT index. Every random access had to start at the first block of the file and track down the FAT table to get to the data block requested. Using DR-DOS to format the disk allowed me to increase the cluster size to 8K while MS-DOS FORMAT only formatted to 2K clusters. This reduced the FAT table access by 4 times and gave an overall performance increase of 3 times.
The CP/M based file systems were archaic but accessing randomly gave a reasonable performance because each directory entry had a larger range of blocks and thus fewer disk accesses.
Unix/Linux INode systems are much better at random access to files than either of those above.
> COMMAND.COM had a lot more internal commands (e.g. COPY, ERASE, support for .BAT files) than the CP/M's command interpreter.
Yes, on an 8086/8088 there was usually much more RAM than on a 64Kb (or less) 8080/Z80 machine so utility code could be incorporated in a larger CCP. CP/M did have a SUBMIT utility though.
> 86-DOS did not freeze when attempting to read/write from an open floppy disk drive.
No, it crashed the program!! On CP/M the BDOS checked that the _correct_ disk was inserted. MS-DOS (at least until 3.1) did not check, if a disk with open files was swapped then MS-DOS would happily overwrite this new disk and corrupt it.
> Concurrent CP/M was CP/M-86 merged with MP/M-86.
This is just dogma. Concurrent-CP/M did not need to be 'merged' with CP/M-86, it was MP/M-86 with added multiple virtual screens replacing MP/M's screen switching code. That is MP/M, MP/M II, MP/M-86 could switch terminal sessions using Ctrl-D. This would hibernate the current program and switch to a different session. Concurrent-CP/M could maintain several sessions per screen without having to hibernate the programs that were in background.
> Both CP/M-86 and MP/M are derivatives of CP/M.
There may well be a case for saying that CP/M-86 is a derivative of CP/M in that the structure of the code of the BDOS and the CCP are much the same and could be regarded as a code enhancement and/or conversion from the original.
MP/M was completely new code. the base code in the system was the interrupt system and the 256 queues. It also had bank switching. The BDOS in CP/M was simplistic and used a single 'thread'. That is once a request was made it ran without interruption until the request was completed. In MP/M and all later multi-xx DRI systems a request was queued and return was made to the system. On appropriate hardware this resulted in the CPU being used for other tasks while the disk was moving the tracks, waiting for rotation and transferring the blocks. At the end of this it signalled that the request was completed and the requesting task was put back on the queue to to scheduled at the next interrupt.
i don't know what you count as 'derivatives'. Certainly all DRI systems had a CP/M-like API, but so did MS-DOS and clones of which there were many. There are even emulators that run CP/M programs on Unix/Linux or Windows.
> DR-DOS inherits from the native x86 version of CP/M which was ported from the 8080.
NO. DR-DOS (and DOS-Plus) derived from Concurrent-DOS, NOTt from CP/M-86. In the original versions it used the same source code tree which was completely separate from CP/M-86.
I have several CONUG newsletters with articles, some written by Gary, that explain this.
> based on the same code,
While CP/M-86 may well be based on some of the same PL/M code for the BDOS, with modifications to handle the segmentation required by the 8086, MP/M, MP/M II, and the various Concurrent-xxx and further derivatives of those were complete rewrites with no common code.
The BIOSes were also completely different code, some of the utilities may have been retained.
Even DR-DOS was a derivative of Concurrent-DOS (with multi-tasking/user removed) and was _not_ a derivative of CP/M-86 with FAT and .EXE support added.
> MS threw together its seat-of-the-pants OS for the Altair
You are very generous in calling it an 'OS'. The stuff written for the Altair was just a BASIC interpreter* and it was loaded into the machine using paper tape created on a DEC machine.
* It is also alleged that the core of the BASIC interpreter was created from an open source BASIC available on DEC machines and used by Bill Gates at Harvard. The math routines needed a complete rewrite for the 8080 written by Monte Davidoff.
> DR-DOS 100% *was* a derivative, yes.
DR-DOS was not a derivative of CP/M, it was a derivative of Concurrent-DOS with (most of) the multi-tasking and multi-user stripped out. C-DOS was a derivative of MP/M-2 via Concurrent-CP/M. MP/M was a completely different set of code from CP/M.
I used (and programmed for) DR-DOS since version 3.0. DRI DOS-Plus (which I have on a BBC Master 512) was also a derivative of Concurrent-DOS and retained some of the multi-tasking. One of the features of DR-DOS was that it would run in ROM because the code segments were 'pure'
> DOS was not a derivative of CP/M
CP/M was developed on DEC machines when Gary was contracted to work for Intel to write a PL/M compiler that ran on DEC to produce 8080 code, so of course it looked like other DEC software.
QDOS (Quick and Dirty OS) was Patterson's original port of CP/M to the 8086 in order to develop an OS for the 8086 version of the SCP Zebra. SCP were full DRI OEM licence hoders with all the CP/M source code that DRI would release (as were Microsoft for the Z80 Softcard that was their major revenue in 1980). The CP/M BDOS was written in PL/M and DRI did not release the source code for that but there were 'annotated decompilers' available for CP/M 1.3 that produced 8080 assembler code from the binary. There was also an 8080 to 8086 translator provided by Intel that would convert the assembler code. It is alleged that this is how QDOS was originally developed. It was brought up on a 8086 Zebra initially by creating a CP/M disk with the QDOS OS on it and then switching processor borads and booting QDOS, thus this did have the CP/M disk format. Later Microsoft provided the FAT disk system from 'Stand-Alone BASIC and this was incorporated to create 86-DOS (otherwise known as SCP-DOS) which was licenced to ibm and MS.
It is alleged that Gary Kildall was able to bring up a DRI copyright message on a PC-DOS 1.0 system and IBM settled with DRI by agreeing to sell CP/M-86 alongside PC-DOS (but never updated from 1.0), gave him money, and rewrote the BDOS to create PC-DOS 1.1 and MS-DOS 1.25.
MicroLAN was 1.25 Mbit/sec. It originated on Cogar desktop machines in 1976. Cogar was bought by Singer Business Machines and then ICL bought part of SBM. ICL produced the Cogar machines as ICL 1500 series (not to be confused with the ICT 1500s). The replacement machines were the ICL DRS20 series which still used 1.25Mbit/sec MicroLAN to connect servers and workstations.
I worked on DRS20, including a prototype, using Retained mode (1500 emulation) and Native mode 8085AH2.
The ICL PC2 (8085 and 8086), Quattro (8086) and Quattro-XM (80286 plus bank switching) were desktop machines with serial terminals running MP/M-2, Concurrent-CP/M, Concurrent-DOS or Concurrent-DOS-XM as appropriate. There was a network system available to connect these computers together using DR-NET software over a token ring by connecting to the synchronous port of the machine. I still have one of the adaptors here. It wasn't MicroLAN though.
An enhanced version of MicroLAN was used to connect DRS303 and DRS305 terminals to DRS300 computers or link computers together. It was still 1.25Mbit.sec.
i still have some of this kit laying around here, a ton or two of it was thrown out a year or so ago.
> business process, ... that might date back decades
The thing about business processes is that they change frequently. Businesses are about competing with others and trying to increase their revenue and profit. They will change the way they set prices, discounts, they will change the product they sell, the way that they sell them. If they don't do that then their competitors will and will take business from them. The computer systems are required to be amended to implement these changes.
> rewriting the code in a modern language
Back in the 80s there was a Datamation survey that reported that 80% 9or so) of COBOL programming was reported as being 'maintenance' whereas 90% of C programming was new applications. Some seemed to conclude that if COBOL programs were converted to C then the 'maintenance overhead' would disappear and the programmers would create new applications.
The whole survey was seriously flawed. The COBOL and C sites were doing completely different types of applications. The COBOL 'maintenance' was, in fact, updating the business applications to handle changes in the business processes as the businesses re-invented themselves to compete better.
> niche languages, all promising to eliminate logic and construct errors by design
I initially read this as:
niche languages, all promising to (eliminate logic) and (construct errors) by design
But I suspect that this should be read as:
niche languages, all promising to eliminate (logic and construct errors) by design
> Or which have most job opportunities?
That would seem to favour languages for which there are the most empty desks. The reason they are not filled may be that few want to use that language.
> how much code is written in each language?
That would favour verbose languages (COBOL, Java) where everything has to be re-invented for every project. Or, possibly, where developer productivity is measured by number of lines of code written per day.
> Google Search and/or tutorials
That would relate popularity of a language to the number of people who don't know it.
It seems to me that Javascript is high on every list because it it hard to avoid when doing web development even when any other language would be preferred.
> the way you've operated a certain business process five years ago is not the way you want to operate it today,
I spent 50 years developing and changing bespoke software for several clients. Businesses are deliberately different in order to be competitive against other businesses in the same market. Not only do they want to be different from others in the market, they want to be different than what they were last year.
The business environment changes: eg banks no longer take cheques, the processes need to adjust, the software needs to follow the processes, or indeed, enforce the changes that have been determined by the management.
On the other hand the business will often operate in spite of the laid down procedures.
> But then, under that (mistaken) belief, so is Unix.
Of course it is. Mac is BSD (Unix) based and it is a great desktop system.
> Taking a server OS and topping it with a window server makes it 'desktop"
Windows NT was developed as a server OS, the original design included it being multi-user*. Windows NT 3.1 had a GUI, basically from Windows 3.1, added on top. The NT design still underlies all the subsequent versions of Window.
> Linux kernel is multiuser server. If [we] press it into single-user desktop use
Unix, Linux and Windows (post 3.11) are all pre-emptive multi-tasking kernels. The only difference in being multi-user is that the user id is attached to the process and this is checked against the various security settings and attributes. As Windows does this too then it is actually a multi-user kernel, it is just not simultaneously multi-user (except TSE and/or Citrix). It can swap users and may even be able to leave the original user with some idle processes until they swap back.
* the multi-user bit was removed because Bill wanted to sell a copy to every user rather than one copy for a bunch of users. This was added back for TSE.
> To say the web was born and grew up on Windows is an outright falsehood
Not only that but Windows 95 initially attempted to replace the internet with Microsoft's own network, the original MSN. To access the internet one needed to buy the Plus pack or use 3rd party software. Later releases did include internet access because MSN failed.
> If you absolutely, positively need performance above all else, then there aren't really many useful alternatives to C
Several years ago I wrote a program in C that merged data into a postscript template, eventually to produce PDFs for such things as invoices and statements. Later, I rewrote it in Python and it ran several times _faster_ than the original.
The reason that the C code was slow was that it used the std library string functions. All the data merging was done using strfind and strcat and these keep scanning along the string to find the null terminator. Python keeps the string length in its object and thus has much faster string handling.
> Linux dead or alive no longer matters for Microsoft.
It obviously _does_ matter to Microsoft, if it didn't they would ignore it. In fact it matters because many cloud customers run Linux workloads.
> Pay for a Windows license, sign in with your Microsoft account and you're free to run Linux.
Or: don't pay for for a Windows license, don't have a Microsoft account and you've always been free to use Linux.
I am not sure why you think that Linux users will replace their systems with Windows running Linux apps. How many Windows users replaced their Windows systems with OS/2 systems because OS/2 could run Windows apps?
In fact quite the opposite happened. Instead of there being OS/2 versions of apps, developers just made Windows versions that would run on Windows and OS/2. Now developers can just make Linux apps that will run on Linux or Windows.
> KDE has been available for Windows for years
So has GTK, the graphical basis of GNOME. Back in the 90s I developed GUI applicationa using Glade* and Python that ran on Linux and Windows with zero code changes, not even a recompile. It even ran on Nokia 800 with a few minor features missing.
* Glade is a graphical interface designer for GTK projects.
Around 1990 I was asked to develop a graphics system to augment an existing text based system. It was for loading container ships and would have graphics representation of the ship and each bay and the containers and of the various tanks. It would have bending and shear graphs, stability, hazardous separations, and much more. It wasn't just GUI dialogs, it was full graphical, it eventually even had a 'helm view' as if looking forward from the helm (or indeed any chosen point) to ensure that the loading and trim met visual sighting requirements.
At the time I was running multiuser systems with Windows 3.1 hosted and an OS/2 system. I had compilers that would produce code for Windows 3.1 and for OS/2 and so I could make a choice whether to do the graphics in Windows or Presentation Manager.
When OS/2 was released with a Windows 3.1 sub-system the choice was obvious.
> Why? would a GNU/Linux developer want to develop any GNU/Linux code on MS Windows anyway?
They don't, but MS wants to force corporates to make these developers use Windows. Clouid users run Linux VMs and developers write Linux programs to run in those VMs. MS wants those developers to use Windows to ensure 100% Windows usage in corporate sites.
Perhaps there will be some tweaks in Azure so that the Linux VMs will only run programs developed with WSL.
> Toshiba notebook which has 32kb of RAM
Most likely it had 32Kb of _Video_ RAM. This number would be displayed when booting. To run any minimal version of Linux would require at least a megabyte, probably 2 or 4 of actual RAM.
Maybe you ran Minix but this required 640Kb.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2023
