Mail Archives: opendos/2002/11/20/08:21:11
Dear DOS fans,
Some of you might be interesting in the following news.
DeviceLogics, Inc., a new start-up founded by Bryan Sparks,
Bryce Burns and Troy Tribe, have acquired DR-DOS from Lineo, Inc.
last month. On 2002-11-18, they announced their plans to release
a new product named DR-DOS 8.0 in 2003. Please have a look at
http://www.drdos.com
http://www.devicelogics.com (probably soon)
for further details.
I would like to take this event as an opportunity to give a little
bit background on this operating system. For those who might not
be aware of it, even though DR-DOS is a genuinely DOS compatible
operating system now, DR-DOS has a long tradition even going
back to the days before Tim Patterson wrote Q-DOS in 1980 and
Seattle Computer Products 86-DOS (1980 - 1981) formed what
then became the basis of Microsoft's MS-DOS and IBM's OEM
version thereof, IBM PC DOS.
Actually, it all started with the late Gary Kildall's CP/M
(Control Program for Microcomputers) for Intel 8080 and
Zilog Z80 CPUs (hence also known as CP/M-80) around 1976
(prototypes even a few years earlier in 1974).
One of the important architectural achievements of the operating
system was the division of the kernel into the lower hardware-
dependent BIOS (Basic Input Output System) and the hardware-
independent BDOS (Basic Disk Operating System) component,
something we would now call a Hardware Abstraction Layer (HAL).
Digital Research's CP/M was portable to many different platforms,
and CP/M programs only using BDOS functions ran on very different
systems, hardware-wise. So, in the Seventies CP/M soon became
/the/ de-facto standard operating system for micro-computers.
When IBM was going to introduce their IBM PC, quite naturally
they also approached Digital Research to provide an operating
system for their machine, but as we all know, Microsoft's then-
new and much cheaper MS-DOS became the default operating system
of choice for the IBM PC as IBM PC DOS, although IBM also
offered Digital Research's own adaptation of CP/M-86 for the
IBM PC in 1981.
At this time, MS-DOS & PC DOS represented basically a rather bad
CP/M-80 clone ported to the 8086 CPU, but with Microsoft's FAT
filesystem and a different buffering logic added. Although the
systems were binary incompatible, one of the primary design goals
for Q-DOS/86-DOS and therefore also for MS-DOS/PC DOS was that
CP/M programs were easily portable to DOS, so there were soon
many traditional CP/M programs available for MS-DOS/PC DOS, and
since transferring programs from the 8080 to the 8086 required
some porting as well even within the CP/M family, there were soon
even more programs available for MS-DOS/PC DOS than for the less
widely known CP/M-86 alternative.
Digital Research's single-user CP/M-86 1.0 - 2.0 (BDOS 2.2,
1980 - 1983) & multi-user multi-tasking multi-processing
MP/M-86 2.x (1980 - 1982) evolved into the single-user
multi-tasking product CP/M-86 Plus (BDOS 3.0???) and the
multi-user multi-tasking product Concurrent CP/M-86 2.0 - 3.1
(aka CCP/M-86, BDOS 3.1, 1983 - 1984), which got an optional
PC DOS 1.1 emulator (PCMODE) added in 1984-09, when the BDOS 3.2
kernel was introduced. The PCMODE module could be retrofitted to
BDOS 3.1 kernels, as, for example, found with one of numerous
OEM versions, CompuPro Concurrent CP/M-86 3.1D.
The first issue of the emulator did only support those old eight-
sector media, so its usability was still very limited - Microsoft/
IBM already had their DOS 2.xx out, which was a significant
rework of the kernel and introduced many of the newer concepts
(dynamically loadable device drivers in CONFIG.SYS, file access
by handles, subdirectories) and already prototyped many of the
internal data structures which still make up DOS today.
At the same time, Digital Research's product was renamed into
Concurrent DOS 3.2 (aka CDOS), which was developed and distributed
in several partially parallel flavors as Concurrent DOS 86
(1984 - ca. 1986), Concurrent DOS 86 XM (a bank-switched
version utilizing an EEMS hardware, ca. 1986 - ca. 1988),
Concurrent DOS 286 (1985 - ca. 1987), and Concurrent DOS 386
(since 1987), each of which was available for many different
OEM hardware platforms. The Concurrent DOS issues for IBM PC
compatible machines were named Concurrent PC DOS instead of
just Concurrent DOS to reflect an IBM compatible XIOS
(Extended Input Output System), although this basic naming
scheme was not maintained all the time.
One of the sidelines (and successor of CP/M-86 Plus) with
a BDOS 3.1 - 4.1 kernel but without the DOS emulator was
the single-user multi-tasking Siemens Personal CP/M-86 1.0 - 2.1
(1984 - 1987), also known as PCP/M-86.
Another sideline worth mentioning was DOS Plus (1.0???) 1.2 - 2.1
(1985 - 1987), which was basically a single-user multi-tasking
issue of Concurrent DOS 4.1 - 5.0 (BDOS 4.1 - 5.0), which already
emulated PC DOS 2.11, so that it ran CP/M-86 as well as native
DOS programs.
At this time, Microsoft and IBM already issued their MS-DOS/
PC DOS 3.0 - 3.2, but DOS 2.xx issues were still in very
widespread use. So, while still laying behind a bit in terms
of DOS compatibility, Digital Research could offer many features
of the technically advanced Concurrent DOS family - features,
which would never be found in MS-DOS/PC DOS.
DOS Plus was available for the Philips :Yes (in ROM), for the
Amstrad/Schneider PC1512, for the Acorn BBC Master 512, and for
the T.R.A.N. Yasmin Turbo - all mainly European products, as
Digital Research's development center for Concurrent DOS was
located in the United Kingdom. (Special cases for some Compaq
and Olivetti machines could also be found in the Amstrad/Schneider
XIOS and other DOS Plus components, but up to this point no
DOS Plus issues for such systems seem to have shown up, so this
might just as well be a left-over from the development cycle.)
Finally, as far as I could research this with the help of several
other people interested in the Digital Research history (see below
for a few links), with the advent of the BDOS 6.0 kernel at the end
of 1987, the famous single-user DR DOS family was carved out of
the Concurrent DOS 6.0 PCMODE and FDOS modules starting 1988-01
by removing the [C]CP/M-86 API layer and replacing CDOS' XIOS with
an IBM and DOS compatible BIOS (DRBIOS.SYS aka IBMBIO.COM).
The first issue of DR DOS, DR DOS 3.31 (1988, BDOS 6.0) emulated
a mixture of the then-current IBM PC DOS 3.3 and the OEM issue
Compaq MS-DOS 3.31. It already supported the new 32-bit sectors-
wide "BIGDOS" partition type 06h for "FAT16B", which was introduced
with Compaq MS-DOS 3.31 to circumvent the nasty 33 Mb partition
size limit. Thereby it was able to access much larger disks than
generic MS-DOS/PC DOS issues.
Also, in some sense, version 3.31 can be considered a "final" stage
of evolvement of the traditional DOS architecture - afterwards most
changes were merely rearrangements of data structures for loadhigh
capabilities with DOS 5.0+, Windows related stuff, or changes for
an improved usability, security, or compatibility with newer hardware,
nothing as fundamental like the redirector interface or such.
So, with the advent of DR DOS, Digital Research had finally caught
up in terms of the DOS compatibility level (although not everything
worked smooth in the early issues, admitted).
DR DOS 3.31 was soon followed by DR DOS 3.32, 3.33, 3.34, and 3.35
(1988, BDOS 6.0), then DR DOS 3.40 (1989) and it really took off
(in particular in Europe and even more so in Germany with huge
OEMs such as Vobis) with the very widespread DR DOS 3.41 (1989,
BDOS 6.3) release.
It is still unclear to me, if the rumored DR DOS 3.42 ever actually
existed - it could well be that what was named 3.42 by some was
nothing but an early Beta version of "Leopard", that is, DR DOS 5.0
(1990, BDOS 6.5), the first DOS on the market to introduce loadhigh
capabilities - a full year before MS-DOS and PC DOS 5.0 incorporated
very similar features in 1991 (with impressively creative play
on words like DOS=HIGH instead of HIDOS=ON, DEVICEHIGH instead
of HIDEVICE, INSTALLHIGH instead of HIINSTALL, or LOADHIGH/LH
instead of HILOAD). ;-)
Digital Research immediately answered with "Buxton" aka DR DOS 6.0
(1991, BDOS 6.7), which built on the DR DOS 5.0 success and added
improved DOS compatibility, even more advanced memory management,
disk compression, task switching, deletion tracking, and the
ViewMAX/2 GUI, a cut down GEM (Graphical Environment Manager,
Digital Research's GUI) derivative.
It took Microsoft and IBM almost two years to respond with their
MS-DOS 6.0 (1993) and PC DOS 6.1 (1993) products.
The original DR DOS 6.0 (with BDOS 6.7 kernel) received a large string
of public updates until 1992. Up to this point, the single-user
DR DOS and the multi-user multi-tasking Concurrent DOS products -
although evolving into very different directions and markets -
were co-developed, partially even from the same source trees.
Meanwhile, the successor of Concurrent DOS 6.2 (BDOS 6.2) had
become Multiuser DOS 5.0 (BDOS 6.5) aka MDOS, and when Novell
discontinued Multiuser DOS 5.1 (BDOS 6.6) in 1992, the development
of MDOS split up into three independent development lines brought
forward by Concurrent Controls, Inc. (http://www.conctrls.com -
their last version was CCI Multiuser DOS 7.22 Gold in 1997),
Datapac Australasia, which developed their Datapac Multiuser DOS 5.0 -
5.1 (BDOS 6.6) into Datapac System Manager 5.0 - 7.0 (1996), which
reportedly is now owned by Citrix, and Intelligent Micro Software
(http://www.imsltd.com), which developed IMS Multiuser DOS 5.0 - 7.0
into the 32-bit Real-Time OS IMS REAL/32 7.50 - 7.93 (2002).
Just recently they released ITERA IMS REAL NG (http://www.realng.com),
which still supports [C]CP/M-86 and DOS applications, but now has
a Linux kernel underneath.
Novell continued to develop the less powerful, but much more
commercially successful single-user DR DOS product line and
introduced the completely revised BDOS 7.0 kernel with "Merlin"
alias NetWare PalmDOS 1.0 in 1992.
PalmDOS was tailored for early DOS palmtop devices with
ROM/Flash file systems, XIP (Execute in Place), a specialized
version of TASKMAX for immediate access to PIM applications,
PCMCIA cards support, and only minimal memory requirements.
It also utilized the patented dynamic idle detection and
automatic active power management.
The development of the PCMCIA stack was inspired by experiences
made with the first two devices of its class, the Atari Portfolio
and Poquet Computer Corporation's Poquet PC, in which's development
one of the engineers, who also participated in the original
standardization process of the PCMCIA initiative, had been
involved previously.
Anyway, BDOS 7.0 was for the first time no longer a heavily
modified CP/M-86 kernel in disguise and coated with a DOS API
emulation, but a natively DOS kernel with genuine DOS compatible
data structures, not only emulations thereof.
This gave a smaller memory footprint and much more compatibility
with many of the dirty DOS applications out there, but some
former features such as no inherent limitation of the path length
for filenames and practically unlimited directory depths (up
to theoretically several ten-thousands levels deep - only limited
by the amount of clusters on the disk and the 64 Kb segment limit
to specify the path), as well as a few neat API extensions for so
called "floating drives" (an implicit way to assign SUBST drives
by just doing a "CHDIR f:=d:\path") had to be given up in favor to
the introduction of a DOS-like CDS (Current Directory Structure) array.
Two only historically relevant projects utilizing the new BDOS 7.0
were never actually released. DR DOS "Panther" (1992) was an issue
of DR DOS with an optional multi-user security module. It also
introduced the unique DPMS (DOS Protected Mode Services) technology
to relocate slightly modified but still perfectly backward compatible
DOS drivers and TSRs into Extended Memory and run them in 16-bit or
32-bit Protected Mode, and KRNL386.SYS (which later became part of
EMM386.EXE 3.00) providing the 32-bit Protected Mode core OS with
memory manager, multithreading, and preemptive multitasking in
virtual machines, and which contained an early implementation of
"true" Windows-like DPMI 0.9/1.0 (DOS Protected Mode Interface).
ViewMAX/3 with moveable and resizable windows was also part of the
Beta distribution.
The other project, DR DOS "StarTrek" in 1992 - 1993, was a team-up
with Apple to let their Intel port of MacOS 7.1 run on top of
DR DOS, store Macintosh resources in the FAT filesystem, and
interface with the new DR DOS 386 multitasker.
A bit later, Novell released the DR DOS 6.0 "business update 1993"
utilizing the new BDOS 7.1 kernel, the very last update to DR DOS 6.0.
To put all these intermediate developings into the global timeline,
it was only now that Microsoft released MS-DOS 6.0, soon followed
by IBM PC DOS 6.1, then MS-DOS 6.20. So far, all the DR DOS
issues still reported themselves as being PC DOS 3.31 (the real
nature of the OS could be retrieved only using INT 21h/AX=4452h
and INT 21h/AX=4451h).
This changed when Novell released Novell DOS 7 in early 1994.
Since this issue had the DOS 5.0+ DOSMGR API implemented, it
could report a more realistic DOS version such as "IBM PC DOS 6.0"
without risking any further compatibility problems with Windows 3.xx.
Since no PC DOS 6.0 actually existed, but IBM PC DOS 6.1 itself
reported as DOS 6.0 internally, Novell DOS 7 (BDOS 7.2) was
sometimes reported as "PC DOS 6.1" by some system-info tools.
Soon later Microsoft released MS-DOS 6.22, followed by
IBM PC DOS 6.3.
Novell DOS 7 publically introduced many of the technologies developed
for DR DOS "Panther" and "StarTrek", including the preemptive
multitasking, DPMI, and DPMS. Instead of the multi-user security
module, however, it shipped with Personal NetWare and network
security. It also introduced code to circumvent the odd but non-
fatal error message created by the famous "AARD" code (in some
issues of Windows 3.1 and other components), which was originally
discovered by Geoff Chappell in 1992 and discussed at lengths
in some of Andrew Schulman's articles and books in fall 1993.
The encrypted "AARD" code was artificially created by Microsoft
for the simple purpose of defending DR DOS by destroying its good
reputation through bad magazine reports by uninformed journalists
and creating rumors among its user base, who were worried about
compatibility and stability with and under Windows. This strategy
was called FUD - Fear, Uncertainty, and Doubt.
After Microsoft's announcement and - much later - introduction
of Windows 95 in late 1995 - which only in their news-speak
no longer ran on, nor had any need for DOS -, and after a long
string of updates (up to update 15.2 in 1996-01), Novell finally
abandoned supporting the product and sold off Novell DOS 7 and
all other remaining Digital Research assets (excluding FlexOS
and X/GEM, which were sold off much earlier) to Caldera in fall
1996, when they approached Novell to license a good DOS issue
to run under DOSEMU for their Linux distribution.
Caldera OpenDOS 7.01 (BDOS 7.2) was released in spring 1997, but
unfortunately it was only based on Novell DOS 7 update 10 or so.
Like much of the older Digital Research stuff, the remaining
updates were lost at that time. Fortunately, after some odyseey
at least most of the NWDOS patches could be revealed again before
the release of the OpenDOS 7.02 Betas in late 1997 and
Caldera DR-DOS 7.02 (BDOS 7.3) in early 1998. (Now for the first
time the product name was written with a hyphen, which caused
quite some debates among users and developers. ;-)
DR-DOS 7.02 introduced many new things, like a special Year 2000
fix for buggy system BIOSes, support for the international date
format according to ISO 8601/DIN EN 28601, an improved INT 13h
handler for non-standard system BIOSes, or a LFN-enabled
COMMAND.COM. Our team also developed the LONGNAME driver to
support VFAT long-filenames under plain DOS and was busy with
DR-WebSpyder, the graphically DOS web browser and corresponding
enhancements to the network driver stack, including a modem dialer.
The 7.02 BIOS and kernel also incorporated the re-engineered
Novell DOS 7 updates and an endless list of other enhancements
(like new relocation and loadhigh options, a much improved
[D]CONFIG.SYS configuration language, an optimized memory foot-
print, and increased speed), of which I am grateful to have had
the opportunity to implement a substantial portion. So, while
the open sourced 7.01 issue was a step back compared to a well
maintained Novell DOS 7 system, DR-DOS 7.02 was again a huge
step forward compared to Novell DOS 7.
Caldera DR-DOS 7.03 was released in fall 1998 with significant
improvements to the memory manager and, again, many other usability
enhancements, Euro currency support, a FAT32-enabled FDISK,
also some fixes for minor problems introduced with DR-DOS 7.02
(but also a number of new problems - sigh)... The last desktop
issue was released in 1999-01.
At this time the company was already called Caldera Thin Clients,
Inc. to distinguish it from Caldera Systems, Inc., which did
the Linux business, and the shell company Caldera, Inc., which
carried out the private antitrust lawsuit Caldera vs. Microsoft.
Then, somewhat unexpected to most who were enthusiastically
involved in the undertaking, the Caldera UK, Ltd. operation
in Andover, UK, where the core DR-DOS and DR-WebSpyder
research & development had taken place, was closed in 1999-02.
The US-based Caldera Thin Clients, Inc. became Lineo, Inc. in
summer 1999 and continued to make cosmetical updates to some of
the docs, but up to present the current desktop issue is still
DR-DOS 7.03 - and AFAIK without any binary changes compared
to the original 1999-01 issue.
Several other DR-DOS issues existed, but they were either internal
technology studies (like DR-DOS "WinGlue"/"WinBolt", which supported
the Windows 4.xx GUI of Windows 95/98/SE, and of which I have
been one of the principal authors in a small R & D team) or were
OEM-only issues like the FAT32/LBA-enabled issues DR-DOS 7.04/7.05
(still with a BDOS 7.3, system files dated 1999-08-19 & 1999-11-30,
now reporting themselves as "IBM PC DOS 7.10"). Reportedly, even
a DR-DOS 7.05/7.06 version (distributed with the most recent
issues of IBM's OEM issue of Ontrack's DiskManager 2000) exists
(which I have yet to see myself, though).
As one of the guys involved in the development of this operating
system, I have to admit, I have somewhat romantic feelings,
whenever it comes to Digital Research and DR-DOS. I guess, it's
the same as with other developers having had an opportunity to
do such an overly creative and satisfactory work in their lifes.
It was a great time and I still have very fond memories of
the calm, smart, and most inspiring people I met. And I think,
I could inspire a few people as well... ;-)
Needless to say, I am very happy to see that DR-DOS is - apparently
again - alive. Actually, I think, DeviceLogics, Inc. should be
headquartered in Phoenix, not Lindon, or so to speak. ;->
It is not clear yet what they actually plan to do, but as an
advocat for the open source idea, I really hope they will finally
make the /right/ decisions for the benefit of DR-DOS as a high-
performance small-footprint operating system and as a continual
technology demonstration, how DOS should have looked like right
from the start, and, of course, to never forget to serve the
still large and astonishingly loyal user community, who helped
with unbroken optimism and many personal resources to keep the
DR-DOS end-user and even much of the supposed-to-be professional
support up and running over all those years when Lineo no longer
did, and without whom DR-DOS would not have undergone such extensive
testing and would probably have been long forgotten and without
much commercial value any more now.
In my opinion, it is very reasonable to target the embedded systems
market (even though more and more companies are migrating from DOS
to Linux for very obvious and good reasons - but DOS is really
"light-weight" compared to Linux and takes much less resources -
in fact, it still runs on an 8088 in 128 Kb of RAM and could be
trimmed down to 64 Kb with some tweaking), so this is approaching
areas, where microcontrollers have their domain today. Depending
on what you want to achieve, one or the other approach may have
unique advantages.
But as a by-product of enhancing DR-DOS for the embedded systems
market, I still hope, DeviceLogics will also bring forward a
version for desktops as well as for Linux and possibly also for
Windows DOS boxes. Or alternatively, that they will at least
release all the sources of those components not needed in
embedded environments under a license like (or very similar
to) the GPL, so that the community could eventually ramp up
and organize the development of these components on their
own behalfs (ideally in cooperation with DeviceLogics and
the FreeDOS project, of course), instead of what has happened
in the past years, when potential open source developers were
condemned to stay idle and bloodly-needed progress on the
DOS front was effectively blocked, unfortunately. At least,
FreeDOS has made some good progress in the meantime.
In the spirit of Digital Research, I would, under reasonable
environmental circumstances, certainly love to put my little
bit into it once again in order to enhance and improve DR-DOS
to become the "ideal" DOS for any purpose, where DOS makes sense...
This way, we could soon have a very mature & flexible DOS platform,
enhanced to cope with many (realistically not all) of the latest
achievements on the hardware front.
Personally, I would also like to see Linux, DR-DOS, FreeDOS, and
PC/GEOS in small 386+ based PDAs similar to the famous, but long
outdated HP 200LX - in a highly rigid housing, with a "feel-good"
mini-keyboard, a crisp high-res gray-scale LCD screen, very
low-power consumption for an extra long battery life, powerful
and convenient PIM applications with mobile e-mail and internet
access, and an open architecture for easy user customarization
and enhancement.
For those interested in further info on DR-DOS and some other
Digital Research related stuff, don't forget to also have a look at:
http://www.drdos.org - Florian Xaver's DR-DOS fan club
http://www.drdos.net - Christoph Fuchs' DR-DOS info
http://www.delorie.com - DJ Delorie's OpenDOS mailing list
http://mpaul.drdos.org - my currently somewhat outdated
DR-DOS pages, but still much stuff
on undocumented features
http://www.yahoogroups.com - "DigitalResearch" forum
http://www.deltasoft.com - Ben Jemmett's GEM site
http://gem.shaneland.co.uk - Shane Couphlan's FreeGEM distribution
http://www.owenrudge.co.uk - Owen Rudge's GEM site
http://www.simpits.org - GEM development list "gem-dev"
http://www.gaby.de - Gaby Chaudry's CP/M stuff & the late
Tim Olmstead's unofficial CP/M library
http://www.seasip.demon.co.uk/index.html - John Elliott's CP/M insights
Not directly DR-DOS related, but possible still interesting for
advanced DOS users and developers:
http://www.freedos.org - The FreeDOS project
http://www.pobox.com/~ralf/files.html - Ralf Brown's Interrupt List
http://www.breadbox.com - PC/GEOS, Ensemble office suite
http://www.tvakatter.org - PC/GEOS, New Deal Office
http://www.arachne.cz - Arachne DOS Webbrowser
http://www.jpsoft.com - 4DOS enhanced command shell
http://www.semware.com - TSE Pro, a powerful DOS editor
http://www.ritlabs.com - Rit Labs' DOS Navigator, a NC clone
http://dn.id.ru/ - Necromancer's DOS Navigator
http://www.nsk.su/~dnosp/ - Open DOS Navigator
Please note, that all this raving is nothing but my immediate
and independent opinion in regard to the latest string of
events. It seems, things are finally moving again - that's
fine... Let's hope, the move will be in the right direction.
Good luck, DR-DOS!
Greetings,
Matthias
PS. If you spot any inaccuracies or omissions, please correct me -
I definitely don't want to spread any rumors.
--
<mailto:Matthias DOT Paul AT post DOT rwth-aachen DOT de>; <mailto:mpaul AT drdos DOT org>
http://www.uni-bonn.de/~uzs180/mpdokeng.html; http://mpaul.drdos.org
"Programs are poems for computers."
- Raw text -