OSDev.org

I believe that they will only be executed in parallel if you use the -j option with make. The remedy is obvious.

If you really want to use -j, include the .NOTPARALLEL pseudo-target.

iansjack wrote:I believe that they will only be executed in parallel if you use the -j option with make. The remedy is obvious.

If you really want to use -j, include the .NOTPARALLEL pseudo-target.

Wow, I didn't know that. I thought they are parallel by default. Sorry.

Greetings
Peter

nexos wrote:It may take me 5 years. Maybe 40. I just want to learn a lot about NT and OSDev in general. And I love huge challenges.

Yeah, good attitude!
That's what I think about my own development. It will probably take years, but that's what I wanna do!

Greetings
Peter

nexos wrote:One big reason I am switching is because after researching Windows 9x, I have noticed a lot of security and stability bugs. The amount of thunking, the concept of so many virtual machines, the close ties to IA32, the use of V8086 mode, the dependence on DOS, and many other things make it very poorly designed.

I think there is a collective relief you've seen the light.

nexos wrote:I think NT is a lot better, and I am a lot more familiar with NT then 9x. So I will clone NT instead. Yes, it is a learning project. It may take me 5 years. Maybe 40. I just want to learn a lot about NT and OSDev in general. And I love huge challenges. It would be difficult to develop under Linux because of VxDs. They are closely tied to Visual Studio.

For reference, the original Windows NT code was something like 4.5 years with an initial 6 head team, growing to 200, and they knew what they were doing already

You'll want to get one of the "Inside Windows" books. The original edition [Custer] covered the original NT 3.1 release published in 1993. Later editions include changes from later Windows versions. I picked up the fifth edition (covering up to Vista) for pennies on ebay.

https://docs.microsoft.com/en-us/sysint ... -internals

BTW, what is the draw to Windows? Is it familiarity with the dev tools on Windows?

Thanks for the book suggestions. I will definitely get those.

kzinti wrote:

thewrongchristian wrote: Windows 95 was run atop DOS much in the same way Windows 3.11 was.

That is false. Windows 95 didn't run on top of DOS.

I should know, I worked on graphics drivers for both Windows 3.11 and Windows 95 in the 90s.

how I fed up with that BS (Win9x run on top of DOS).

thewrongchristian wrote: BTW, what is the draw to Windows? Is it familiarity with the dev tools on Windows?

BTW, don't you know why such questions are so selective and only appear if one proclaims his/her plans to clone Windows? why they are absent directed at sh1tloads of UNIX/Linux clones being tried to be developed here? funny, but the author is making a UNIX clone too! still no wondering about that from you. do you think it's impossible to want to clone Windows? only UNIX can be "attractive"? just wondering for wondering.

zaval wrote:BTW, don't you know why such questions are so selective and only appear if one proclaims his/her plans to clone Windows? why they are absent directed at sh1tloads of UNIX/Linux clones being tried to be developed here? funny, but the author is making a UNIX clone too! still no wondering about that from you. do you think it's impossible to want to clone Windows? only UNIX can be "attractive"? just wondering for wondering.

Well, because if you write a Unixoid system you have components open-source ready at hand, like X Window system, Gnome, systemd, bash and more! You can concentrate on the kernel and some stuff you like (could be shell or C-Lib or whatever you like).

If you write a Windows clone, you don't have such components. You have to do it all yourself: Kernel, Shell/Explorer, GDI, Unicode, OLE/COM, Registry etc.

Greetings
Peter

zaval wrote:

kzinti wrote:

thewrongchristian wrote: Windows 95 was run atop DOS much in the same way Windows 3.11 was.

That is false. Windows 95 didn't run on top of DOS.

I should know, I worked on graphics drivers for both Windows 3.11 and Windows 95 in the 90s.

how I fed up with that BS (Win9x run on top of DOS).

Well, DOS was REQUIRED to bootstrap Win95. The fact Windows replaced much of the file access once up and running with it's own 32-bit protected mode implementation (I bet it was compiled from the same source as the real mode DOS version) is neither here nor there. Windows for Workgroups used a similar mechanism to Win95, backported from the then Chicago project before Win95 was released, for much the same reasons, but nobody would doubt that Windows 3.11 was tied to and ran on top of DOS.

Win95 could also access devices that only had a DOS driver. It would have had to do that via the real mode component of DOS, and it had to keep the real mode data structures (presumably filesystem information) in sync with the protected mode equivalents.

More to the point, Microsoft charged for them as separate products, which they couldn't if Win95 was truly stand-alone. A court even ruled the two were tied together illegally.

So, from a product, marketing, legal if not technical basis, Win95 was intimitely tied to and dependent upon DOS.

zaval wrote:

thewrongchristian wrote: BTW, what is the draw to Windows? Is it familiarity with the dev tools on Windows?

BTW, don't you know why such questions are so selective and only appear if one proclaims his/her plans to clone Windows? why they are absent directed at sh1tloads of UNIX/Linux clones being tried to be developed here? funny, but the author is making a UNIX clone too! still no wondering about that from you. do you think it's impossible to want to clone Windows? only UNIX can be "attractive"? just wondering for wondering.

I just find cloning Windows a strange choice. It's not a very clean OS (or at least, Win32 is not.), and especially the original choice of cloning Win95 is bizarre given its architectural, erm, challenges. Also seems a strange choice to have two separate clones, when one could be the base of the other and have a single code base.

If you want to learn how OSes work, or how to write one, there are easier, more accessible targets. My own OS will be more UNIX like too, and if you asked me why, I'd be happy to answer.

Windows 9x had DOS for two purposes: bootloader and legacy driver compatibility layer. Windows 9x was dependent on DOS, but not at all like Windows 3.x. I did choose not to clone Windows 9x, and am still on the fence for cloning Windows, 9x or NT, period. I think modern OSes are large, slow, and bloated. It takes my Windows PC 5 minutes to log in! Plus, it is new with Windows 10. I read about the many Linux codebase problems today as well. I think a new architecture still may be a good choice, and am considering that instead of a Windows clone.

nexos wrote:I read about the many Linux codebase problems today as well.

What problems?

I think a new architecture still may be a good choice, and am considering that instead of a Windows clone.

That seems like you are a bit like the prototype "Mister Perfect"

Greetings
Peter

Linus himself stated that he had envisioned a "small, streamlined hyper-efficient kernel" but that it was now "huge and bloated". Kernel developers said that bugs get swept under the rug because the kernel is so bloated they can't find the problem. I've Iooked through the directory structure of the source and thought, "my kernel has better structure then that!". I'm sure similar problems bug Windows too, as windows is very bloated. Two functions for everything that has to do with strings is way overkill. I think I will put my name with Mister Perfect now that you mention it .

the IP of my ISP got blacklisted* (again) by that spamhouse thing, so I couldn't post answers timely. here is late version. hopefully it will work.

* - or, should I say now whitelisted instead? given all this idiocy with terms renaming.

PeterX wrote: Well, because if you write a Unixoid system you have components open-source ready at hand

and you have open source kernels as well, - linux/minix/bsds etc, so you don't need to do anything at all.
or you can implement a POSIX subsystem for your NT-like OS and run those posix programs.
or maybe it's not about practicality but about what you do like to do?

thewrongchristian wrote: So, from a product, marketing, legal if not technical basis, Win95 was intimitely tied to and dependent upon DOS.

whatever. you have been told by the knowledgeable person on this topic above.

I just find cloning Windows a strange choice. It's not a very clean OS (or at least, Win32 is not.), and especially the original choice of cloning Win95 is bizarre given its architectural, erm, challenges. Also seems a strange choice to have two separate clones, when one could be the base of the other and have a single code base.

Nothing strange. it's only for you. because "not a very clean OS" - is an extreme bias, without anything convincing behind it. I oppositely think WinAPI is way cleaner, and much better, powerful and thus attractive than posix.

If you want to learn how OSes work, or how to write one, there are easier, more accessible targets.

the problem with this statement is that it's a biased, subjective opinion of a unixoid, whose religion calls on to have a low opinion on all Windows, no matter what, whereas it has been given as something general, objective, trustworthy. It's just if I came up and said to one of those parroting linux: "oh come on, linux is an awful calque itself, there are much better sources to learn how OSes work". but why should I get into someone's preferences? do I really need to force my preferences on others? do I hope to convince anybody? for what? also, it's not about easier. it's about what you want. so when someone "just wondering" about others' choice in this situation, it looks more like a hidden attempt to express one's own irritance, like you know "hold on, dear, there is someone wrong in the Internet" phenomenon.

My own OS will be more UNIX like too, and if you asked me why, I'd be happy to answer.

surprise! no thanks, I know the answer - it's a kind of religion - worshipping unix. no real reasons really.

Sorry to hear about your internet connectivity problem.

zaval wrote:

PeterX wrote: Well, because if you write a Unixoid system you have components open-source ready at hand

and you have open source kernels as well, - linux/minix/bsds etc, so you don't need to do anything at all.
or you can implement a POSIX subsystem for your NT-like OS and run those posix programs.

Yeah, you could build a Windows-clone-kernel and POSIX-tools on it. And you could build a Unix/Linux kernel and Windows-tools on it.
Both would reduce the amount of work. But both options aren't popular, are they?
I would like to see such systems (both kinds).

PeterX wrote:Sorry to hear about your internet connectivity problem.
Yeah, you could build a Windows-clone-kernel and POSIX-tools on it. And you could build a Unix/Linux kernel and Windows-tools on it.
...
I would like to see such systems (both kinds).

It's hard to say if they are popular or not. there are two worlds - Windows and Unix clones. both worlds not surprisingly tend to use their own stuff and being skeptical about each other. but this is about fans. most people use (and program) what's available and what has won its place over there. desktop is fully dominated by Windows, and as WinAPI and all the galaxy of satellites provide everything, why would you need something else? I don't see, what POSIX or any linux specific thing could add to that.

every program worth using, runs on Windows.

POSIX is much limited, but seems enough for Unix clones where they are mostly used - web servers.

talking about a POSIX subsystem in an NT-like OS, I meant NT is a modular architecture. it's comprised of a set of separate components. so it's easier to introduce POSIX support by providing some additional components - "support" processes, that would create a UNIX view of machine for programs, DLLs loading into every such program and supplying expected POSIX functionality and serving as a gateway into the kernel part, that would optimise POSIX system services (calls) not needed for anything else. you don't need to redesign your monolithic mess to adapt yet one thingy, you don't even need to recompile your kernAl! you need to write some "package" of programs, libraries and drivers and voila - now your NT clone is at least source level compatible with a pile of available POSIX programs.

Both would reduce the amount of work. But both options aren't popular, are they?

WSL, WSL2? Whine?

zaval wrote: every program worth using, runs on Windows.

And Mac OS. Nobody has mentioned making an OS that can run Mac programs