On Wednesday 17 July 2013 10:39, DecadentLinuxUserNumeroUno conveyed the
following to alt.os.linux.ubuntu...
Post by DecadentLinuxUserNumeroUnoOn Wed, 17 Jul 2013 06:21:57 +0200, Aragorn
Post by AragornOn Wednesday 17 July 2013 04:53, DecadentLinuxUserNumeroUno conveyed
the following to alt.os.linux.ubuntu...
On Tue, 16 Jul 2013 19:05:54 +0000 (UTC), Dustin
Post by DustinActually, they are.
Actually, you were misinformed. Then you took it as gospel.
Any REAL computer science education behind your claims?
Uh, DLUNU, he /is/ right. See my elaboration elsewhere. They /were/
indeed DOS-based, but running as DPMI (DOS Protected Mode Interface)
with a bolted-on task scheduler and the Windows 4.0 graphical user
interface.
You're right. It was a GUI on a "DOS" framework, as was previous
Windows versions. The first one without was Win2k, right?
Well, no, it's more complicated than that. You had the NT-based Windows
family on the one hand and you had the DOS-based Windows family on the
other hand.
Both were marketed side by side, but the first Windows generation where
they were not marketed side by side anymore was Windows XP, because when
Windows 2000 (NT 5.0) was being marketed, Microsoft was still selling
Windows Millenium Edition, which was a continuation of Windows 9x.
So, Windows XP was the first consumer-based Windows version which ran
off an NT kernel instead of off a DPMI base.
Post by DecadentLinuxUserNumeroUnoThat "DOS" version, however, was not the same as the previous (as in
an "MS-DOS" one could buy stand alone, and by rights should have some
moniker included when mentioning it that keeps lay folks reading such
a discussion as this to understand that it is not nor was never
intended to be called "a DOS" separate from the declared OS name
"Windows" by that time.
One could indeed not purchase a separate license for the DOS in Windows
9x and ME anymore, but that was Microsoft's intent all along. See, by
integrating DOS into Windows 9x/ME, Microsoft could prevent that people
would be using DR DOS, Novell DOS or any of the other standalone DOS
versions in combination with Windows.
Post by DecadentLinuxUserNumeroUno[...]
I do not recall many folks running things from the command line in
those days where they were not doing it from an older, true "DOS"
(MS-DOS actually). So it really is just semantics and the need to
make sure that in our discussions, we illuminate readers where the
breakpoint was between hard DOS users and the new stub (hehehe) that
mounts the new, 32 bit windows memory management schema.
Well, the first commercially viable graphical user interface was of
course that of the Apple MacIntosh machines, and both Microsoft and IBM
were well aware of that.
As MS-DOS/PC-DOS could not make full use of the extra power of the Intel
80286 and later processors, IBM took the initiative of developing a
successor to MS-DOS/PC-DOS, which would be backward-compatible with DOS
and which would maintain the look & feel of DOS, but with protected mode
operation and multitasking. Because of their cooperation with Microsoft
on DOS, IBM involved Microsoft in the development of this new system,
and this was when OS/2 was born, but the first version of OS/2, version
1.0, did not have an actual graphical user interface. It could display
applications in windows on the screen, but it was still in character
mode. The next version however, OS/2 1.1, made use of true pixel-
oriented graphics mode.
Microsoft had already released early versions of Windows, which also
more or less used pixel-oriented graphics, but didn't really offer
anything more, and as such, Bill Gates decided to give Windows 3.x the
look & feel of OS/2, and to give it multitasking abilities, but all
while still based upon MS-DOS.
Now, OS/2 1.x was developed for the Intel 80286 processor, and Intel had
already released the 80386 and shortly after that, the 80486. Those
were 32-bit processors. IBM and Microsoft then came to an agreement
that IBM would be developing the 2.x generation of OS/2, and that
Microsoft would be developing the 3.x NT generation of OS/2, which would
share the same 32-bit underpinnings as Microsoft's own Windows 3.x NT.
These underpinnings were to be the Mach 2.5 kernel.
However, due to the success of Windows 3.x for DOS, which came pre-
installed /with/ DOS on many brand-name computers, Bill Gates decided to
focus his efforts on Windows only, and he broke the agreement with IBM
for the development of OS/2 3.x NT. Also, instead of going with the
UNIX-style Mach kernel, he hired Dave Cutler at DEC to write the NT
kernel on a VMS base - Cutler had co-written the VMS kernel and he hated
UNIX.
As such, the foundations for Windows NT had been cemented, but NT was
definitely a work in progress, because Gates wanted to be as compatible
with the DOS-based Windows 3.x versions as possible, whereas the NT
kernel was a VMS-clone and thus quite a different thing. This is why
Gates decided to continue marketing and further developing DOS-based
Windows versions for the consumer market until NT was fully ready to
take over.
Gates's idea was to have developers use the Windows-native API of
course, rather than have them continue to develop applications for MS-
DOS. But DOS allowed developers to pull all sorts of tricks due to its
primitive nature and the fact that it offered direct access to the
hardware, so not all developers were eager to jump on the Win API
bandwagon, and due to the DOS-based nature of Windows 9x and ME, even
those developers who did write for the Win32 API still amply made use of
the underlying DOS layer, which enabled them to do things that the NT-
based version of Windows would not allow.
Post by DecadentLinuxUserNumeroUnoI do not recall booting to it when I had that version of Windows,
but I cannot say for sure. I may well have. I used to do things in
labs back then. Gots lots better gear these days. Electronics has
been good to us all. Most "advanced lab gear", even recent models run
Windows 2000 as their underlying control OS. Maybe it is embedded.
more likely flash or other stored media method. Embedded would not be
updateable as easily. They'll probably move to Win 8 next. Damn.
nice to see Redhat or something the businesses trust in there. Kinda
like what cisco did.with their routers and switches.
Well, Linux is by far the most prevalent embedded system these days, and
then there's also QNX, which is also a UNIX variant. I don't think
Windows 2000 was ever used as an embedded system, but it was definitely
more popular for enterprise-grade stuff than the later Windows versions.
Enterprises don't upgrade as often as desktop users do, and Windows 2000
was considered more stable/reliable than NT 4.0, whereas later Windows
versions such as XP and Vista (and their respective 2003 and 2008 Server
versions) were perceived as less reliable and more bloated.
Post by DecadentLinuxUserNumeroUnoPost by AragornAlmost all of the I/O even ran through the real mode DOS drivers,
[...
One could still install things in config.sys as well, IIRC during
the boot process, prior to Windows popping up (a full minute later)
:-(.
Yes, that is correct.
Post by DecadentLinuxUserNumeroUnoPost by Aragorn...] via the /virtual device drivers/ of Windows itself, which
presented virtual device driver APIs to the Windows applications, but
served as an abstraction layer for the DOS device drivers, which
themselves were still running in real mode and using legacy real mode
BIOS calls.
Which is where he (Bill) formed his ideas about the NT HAL he
subsequently had engineered into what became NT. That was "New
Technology".
Well, no, NT and HAL are already a lot older - the first NT release was
3.1 and appeared in 1993 - and the ideas for the HAL came from VMS. And
the NT moniker was something which was destined from the start to be the
future 3.x generations of both Windows and OS/2, back when IBM and
Microsoft were still working together. NT was even still supposed to be
based on a UNIX-style Mach kernel back, rather than on a clone of the
VMS kernel as it is now.
Post by DecadentLinuxUserNumeroUnoI remember running Quarterdeck memory management when I ran
DesqViewX That managed things as both expanded and extended memory.
DesqViewX had to have it to run. I think it utilized protected mode
as well.
Well, yes and no. DesqView itself did not use protected mode. It used
EMS (Expanded Memory Specification) via Quarterdeck's QEMM, but QEMM
itself was a technology which made use of protected mode to simulate
real mode expanded memory.
To cut a long story short, EMS provided for a "movable memory address
window", and when either QEMM or the DOS-native EMM386.SYS/EMM386.EXE
drivers were used as the memory manager, the memory addresses this
window would point to were situated in what would normally be the
protected mode address range. So the processor had to switch to
protected mode to access that memory, and then remap it to a real mode
location within the EMS window, and then switch back to real mode so
that DOS could use those memory contents.
A similar technology was XMS (eXtended Memory Specification), which,
instead of using a movable address window, would simply copy 64 KiB
segments back and forth between the real mode memory range and the
protected mode memory range. To DOS, it was in essence a similar
technology to swapping, except that instead of storing a swapped out
segment on a slow disk, it would store it in protected mode memory.
Post by DecadentLinuxUserNumeroUnoPost by AragornWindows 3.x, 9x and ME spent approximately 60% of their CPU time in
real mode, and the remaining 40% was running in ring 0 of protected
mode, even for application-level software. There was no privilege
separation,
And therein lies the problem. He (Bill) could never get it through
his head that one cannot allow a crash to run down into that lev...
As I recall, he referred to the crashes afterwards as "a feature", and
neither he personally nor Microsoft as a company ever took crashes
seriously. They consider crashes to be "acceptable behavior".
Post by DecadentLinuxUserNumeroUnoPost by Aragornand everything running in protected mode all ran within the same
address space. Any Windows application could (and often did) crash
the entire system.
Yet another reason he should have done a full rewrite from early on.
but noooooo. Took him a while. Things moved slower then too though
guys... circuit board form factors... hardware layer stuff, and the
glue. And he had money to make and projected schedules keep track
with and accelerate (or insure endurance of) current sales to clear
inventories. He had discs to reproduce and sell! What a life!
Post by AragornDustin is right, and Virus Guy is the one who was wrong.
I am wrong a lot too, but I was there, and understand the position.
not exactly just semantics but I do not recall ever calling it a DOS,
as in the previous moniker, not the literal acronym. Of course they
are all that.
So, "the stub" remains and doesn't get kicked out, But is there
then a "quit to DOS" "feature"?
Yes, it was possible to quit the Windows session and fall back to DOS.
And it was also possible to boot up in DOS-only mode. Usually, the
installed versions of Windows 9x and ME booted straight into the Windows
GUI, but it was possible to get it to boot into DOS only by way of a
shortcut key - F8, if I'm not mistaken - or to even create a boot-up
menu, which was called from CONFIG.SYS and which worked in conjunction
with AUTOEXEC.BAT
--
= Aragorn =
GNU/Linux user #223157 - http://www.linuxcounter.net