Can tell anyone why Windows 9X discontinued and just replace home (consumer) edition of Windows NT series?

1. Obsolete
2. Outdated
3. Unreliable

1. Maintenance of a second code base is nonsense unless the functionality is different. The NT platform, by 2000, was almost entirely compatible with Windows 98. Which one should MS choose? This is I think the most fundamental problem with the 98 code base. There was no advantage.
2. Worse performance compared to Windows 2000, especially for IO see http://books.google.com/books?id=iiuZnC ... e&q&f=true .
3. Windows 98 was unable to support multiple CPUs

1. It's slow
2. It's Unstable
3. It's outdated
4. It cannot handle modern hardware and software
5. The NT line was actually better for everybody rather than the 9x/DOS

Seriously, I know you knew the reasons why it was replaced. I know this is just a way to boost your post's.

In start windows nt didnt had support for all computers.I tried to install windows nt 5 beta but nothing.Now we are using Windows 7(most of us) and waiting for windows 8 which will have support for tablets

All of them based on the NT core.
Basicly, the development of the 9x line wasn't worth it, because it would get more unstable from time to time, i.e. see what happened with Windows ME, it bluescreened every single time you right-clicked, at least in a Pentium 3 Katmai processor. Don't even mind installing a driver, that was an impossible task without the ability to right-click.

Anyways, Windows 9X was discontiuned for it's unstability and I know some folks want's to have 9x as thier primary OS. Windows 9X runs great on Pentium II and III and in my retro heart, Windows 9x shall never be discontiuned and it shall live for ever.

Windows 9x/Me was discounted because:
1. it's unstability.
2. it's slow.
3. it's old.
4. NT series was very, very better than 9x series.

Wow, there's a lot of rubbish in this thread.

Not worthy of an argument, however, as the NT-lovers will always repeat garbage about 9x and us 9x-lovers are not interested, and disagree 100%.

9X ran on top of the aging DOS platform and was less stable than NT. For example BSODs can be quite common in 9X while they are very rare in NT. ME was a pretty bad edition of Windows that finished off 9X for good, while at the same time 2000 based on NT was one of the leanest and most stable versions of Windows. Look at how much success XP has had which is based off 2000, even 10+ years later it is still widely used.

NT is better. Get on with it.

That NT is better is subjective. While I admire NT for having a level of stability 9x never had, I still like the backwards compatibility 9x has. Specifically, real DOS in 9x is a great advantage it has over NT. Not just for compatibility itself but also because it helps fixing the OS if something goes wrong. In NT, if you can't boot into the GUI or you can boot but you have a piece of malware that's preventing you from fixing the OS from the GUI, you have to rely on third party boot CD's and so on to get access to the files. If the same happens in 9x, you just boot into DOS, find what's wrong, fix it and boot into the GUI again.
In this, 9x is actually closer to *nix - *nix is effectively a GUI running on top of a text mode OS and so is 9x. NT on the other hand has the GUI in its core. I feel that if the guys over at Microsoft had bothered rewriting 9x so that it retains the DOS-based kernel but gains the NT level of stability and features, it could have become better than NT.
Maybe the solution would have been to add a 32-bit DOS and make the 32-bit Windows kernel load on top of that, but just keep the boot loader 16-bit (which it is in NT as well) and make it able to load real DOS if the user requests so by enabling the code that calls the actual DOS kernel while disabling that code if the 32-bit OS itself (whether in 32-bit DOS mode or 32-bit Windows mode) is requested to load.

So the way development turned out, NT came out better than 9x but had development gone another way, 9x might have become better than NT. It's also worth noting however that Microsoft have planned to marge 9x and NT into a single NT line for ages before finally succeeding with it. Originally, Windows 3.x was supposed to be the last Windows version of the old line and Windows 4.x was supposed to be the NT-based Cairo, if I read the Antitrust documents right.

Rubbish.
I'm not interested in your opinion, and obviously you're not interested in mine. So let's not waste each other's time.


+1


Yes. Windows 9x was never supposed to happen. NT was Microsoft's baby, but it was Windows 9x that helped them "take over the world." Therefore they hate it, because it did for them what their pet NT never could.

Now, all you 9x-haters take that and "get on with it."

9x was good enough back in it's day. NT is just superior today.

Also, DOSBox.

I don't even...

NT is better.

That is not open to debate. There are a number of ways to determine or measure the technical ability of a kernel, and NT surpasses 9x in every single way.

9x was never anything more than a dirty little hack to breathe life into aging code whilst they refined the real deal - NT. 9x is structurally unsound at best, an example of gross negligence in software engineering at worst.



There's no such thing as a "text-mode OS". It might be that an OS choses to output to a console by default rather than a GUI, but this does not make it a "text-mode OS". The NT model with GUIs is superior here because it treats text-mode and GUI for what they are - equally valid output options. Just because there's a preference for Windows to use a GUI (hardly a terrible decision) does not change anything.

DOS is a dead-end. It's a giant mash of legacy, x86-specific code. It's not that it used x86 ASM - I don't know how much of it was x86 ASM by the end, I think at least some of it was C by the end - but the assumptions it made about the hardware. It was rigid to a classical PC model. That had no future. NT is not like this - NT makes no assumptions whatsoever about hardware. It is as portable as any other OS out there - any platform specific stuff is confined to the HAL.

The NT bootloader remained 16-bit (on x86 only) for so long because that was a limitation of the BIOS. Thankfully, this terrible legacy mess is finally being dealt with in [U]EFI - If you have an EFI-compatible system, then the bootloader will be the same as the rest of Windows - on x64 systems, it's pure native x64, and same for the other architectures (ia64, x86 in Windows 8).

Your suggestion of using the DOS kernel must surely be a joke. I can assure you now, for the DOS kernel to work on say, SoC or ARM, it'd need a complete rewrite. We could maybe use NT as a starting point - owait...

I speak from my own personal experience here when I say NT > 9x. On NT, I very rarely see any STOP errors and when I do, it's because I have bad drivers, failing hardware or am running a pre-release version. Back when I used 9x, it was quite the different story. I would get crashes much more often and a lot of times they would lead to a BSOD. Even a regular app crash would lead to hard lockups, system instability and BSODs. So as far as stability is concerned, NT is the superior OS.

Also, who writes programs for Windows 9x anymore? Answer: No one. The people who still use 9x don't make up enough of the market share for developers to bother with these days. True, stuff like KernelEx exists but that only ports a few Win2k/XP APIs back to 9x and soon enough, developers will drop Win2k/XP support.

Let's also not forget that 9x is not multiprocessor aware. Got a system with two CPUs? A multi-core CPU? Win9x may run on them, but it will only recognize one CPU/core. Oh yeah, 9x is also 32-bit only so in theory, all you can access is 3.5GB of RAM. I don't think 9x ever had PAE as well.

All that said, 9x was good enough for it's time. I still have one machine in my whole network still running Windows 95. Great little 486-based system for running DOS apps on.

Of course NT can be considered more "modern" and more "up to date" because it's development has continued, whereas 9x development has ceased. This is comparing apples and oranges. But the fact is, Microsoft and NT would not be what they are today without the popularity of Windows 9x.

You're all welcome to your opinions. If you like NT better, then more power to you. I'm not here to change your minds. But rest assured, you will not change mine. When Windows 9x will no longer do the things I need it to do, then I will use Linux.

I could go on and on about how meh Linux is (despite my username, lol immaturity) but I'm not going to. At least you acknowledge we all have our own opinions and you aren't trying to force your own opinions on us, unlike some people I'll keep nameless.

NT is more stable and more secure. But the backwards compatibility is better in 9x.


Sadly this here is true. However that's purely due to the development strategy Microsoft took - namely, gradually make 9x more and more compatible with NT until eventually it's compatible enough that it can just go away for only NT to stay. And it seems to have worked.


Well I was talking about user interface. Windows 9x has a text user interface at its core, with the GUI, while starting up automatically with the OS, being an optional addon. You'll find *nix work pretty much the same way. And it does change things - in NT, I need to run the GUI in order to run the text-mode command prompt which outputs in text mode and as such shouldn't need the GUI in the first place.


DOS is pretty portable. I see versions for the DEC Rainbow, NEC PC-98, Fujitsu FM-Towns and so on. And they all had very different hardware, only sharing the same x86-based CPU however with different BIOS'es and interrupts (for example, I seriously doubt the PC-98's interrupt input/output interface is compatible with the INT 10h input/output interface from a regular PC, for example). Running PC-98 DOS on a regular PC will crash, proving how different it is.
DOS is pretty much a very basic OS and as such has much less to be ported to a non-x86 CPU than Windows of any line has.


Yeah, if one has an EFI-compatible system, and apart from Apple machines, EFI is still the minority.


It would need a complete port of the kernel and a simple recompile of everything else.
Also, what I proposed is to relegate the original, 16-bit DOS to an optional component that isn't even loaded if the 32-/64-bit OS portion is loaded. Basically an OS with 3-levels:
- 16-bit boot loader for PC's with BIOS with optional 16-bit real DOS [note that the 16-bit boot loader wouldn't even be used on a system with EFI];
- 32-/64-bit DOS of the same architecture as the Windows and of course be a complete rewrite from scratch but just using the same user interface of 16-bit DOS;
- 32-/64-bit Windows running on top of that rewritten-from-scratch 32-/64-bit DOS.

Of course that would no longer be 9x but something completely new that would take ideas from both 9x and NT and combined them into one powerful OS.

Of course, then you have to do a whole bunch of extra work to port all of your applications over to non IBM PC platforms with DOS. NT applications also require porting to different architectures, but as far as I understand, it's just a matter of throwing the source code through the proper compiler instead of having to rewrite basically the whole application.


True, but EFI is quickly becoming more and more popular. My new laptop came with EFI; I suspect that OEMs have been shifting over to EFI pretty recently.

Yes, EFI is finally seeing some widespread adoption since with Windows 8, it'll be required for a PC/motherboard to get the Windows certified logo. Any current generation tech from the mid-range up will definitely have UEFI these days, even if it still looks and feels like a BIOS (some do). Indeed, I believe that AMD has already mandated UEFI with AM3+ motherboards - I've certainly not seen one without UEFI.



The User Interface is not part of the kernel - it's an OS-level design choice. There's nothing to stop NT being text-mode - indeed, you'll find in XP, a number of examples of it running in Text Mode, like the first stage of the setup (which was NT, not DOS as is often claimed). It's a simple fact that Microsoft have made a conscious choice to make as much of the OS GUI as possible, and this would have happened regardless of the kernel. My point was that unlike in DOS and *nix, GUI is not a second-class citizen in NT, it's equal to text-mode in that the kernel just considers it another way to output stuff. This is the better option from a technical standpoint.

DOS is not as portable. You mention alternative architectures, but they are still incredibly similar compared to say, an amd64 PC and an ARM SoC. NT's portable architecture means that such a smaller portion needs to be changed. With DOS, it's rewriting the kernel, as you admit. With NT, it's just a new HAL, which is a tiny fraction of the code.

There's no need for this DOS layer you propose. NT already handles this with far more robustness and flexibility. Adding DOS-like parts to it would weaken the concept, not improve it.

People, take a look at how far NT has gone and say 9x can go the same length without rewriting every part of the kernel. NT is no different to 9x for backward compatibility, in fact NT is superior in this respect. You all seem to forget NT4 and it's predecessor, the NT 3.xx series (NT3.1->NT3.51). They were indeed around in the nineties too! To be brutally honest though, who uses software from the 90s on a general scale? Note the phrase general scale. This does not include your abandonware testing adventure.

I really cannot see the point in this.
Separation between OS and GUI is good, but I think it's awkward to say 9x is someway closer to *nix just because of that.
*nix is good 'cause OS and GUI are separated each other AND that OS (even if you're on CLI) works damn well, and sorry, i can't say the same for DOS.
And honestly the malware example doesn't work for me, NT safe mode is quite good (and it can provide you with a command prompt) when it comes to problem fixing and anyway in some cases even DOS would be useless.
Keeping OS and GUI separated is more a matter of design, it simplifies things a lot, but if you plan on developing a modern OS based on DOS you have a serious flaw in your design, way worse than if you put GUI in the kernel.

He's saying that the design paradigm is closer to *nix in terms that its main user interface is that of a command line. What's so hard to see about that? I also wouldn't be so sure about pointing out DOS's vulnerabilities because mind you, even Microsoft's *nix server have gotten infected. However let's be honest, DOS is an aging platform and it must be rewritten if anyone expects it to be used for a modern day OS.