OSDev.org

Just came across some weeks about ReactOS and it s NTOSKRNL knockoff of the NT Kernel and installed it and works really fine even faster than my installed Mint, so i was courius if there some OS using it as an Base (like carnavigation or blackberry os use qnx kernel) but there is really nothing using at as base so why?
Whats youre Expierence with it?
Im thinking at the moment that a port to mobile phones is a good idea since it support s ARM
So my pros of the kernel

-Fast
-Not Linux(Everything uses Linux monocultur some variability would be nice in my opinion)
-Microkernel(Easy to port)

cons
-legal questions
-drivers limited?
-must be more...

Cons:
- A crazy hobby project.
- Limited support for Windows software. No UNIX-compatibility (AFAIK).
- Bugs.
- Small community (compared to Linux and *BSDs).

Personally, I see ReactOS as a monumental and interesting project.

1. It's a complete (API compatible) reimplementation of Windows (adhering to published APIs, and probably has close to full binary compat too. This means windows drivers should work)
2. Legal issues were addressed a few years ago (the code is legal, being a clean-room reimplementation)

However, it does suffer from being in an eternal state of catchup with windows. BUT I forsee it filling a similar role to FreeDOS in the future, allowing legacy software to continue to be used. Presumably, ReactOS could take some different paths wrt handling legacy software to maintain security while still running Win9X sofware.

Hi,

thepowersgang wrote:Personally, I see ReactOS as a monumental and interesting project.

I agree - it's an impressive achievement (especially considering how convoluted and undocumented the Windows kernel is).

dstyl wrote:Just came across some weeks about ReactOS and it s NTOSKRNL knockoff of the NT Kernel and installed it and works really fine even faster than my installed Mint, so i was courius if there some OS using it as an Base (like carnavigation or blackberry os use qnx kernel) but there is really nothing using at as base so why?

The kernel is designed specifically for a Windows clone; which means that it's not suitable or ideal for anything else. If you're planning to write a Windows clone then it'd be good choice for a kernel, but if you're planning to write a Windows clone (e.g. using the ReactOS kernel and attempting to compete against both ReactOS and Windows) then it's a massive waste of time and you should just join the ReactOS project instead.

Basically, using this kernel as a base for your own OS project would either be extremely silly (your OS isn't a Windows clone) or extremely silly (your OS is a Windows clone).


Cheers,

Brendan

Basically, using this kernel as a base for your own OS project would either be extremely silly (your OS isn't a Windows clone) or extremely silly (your OS is a Windows clone).


Cheers,

Brendan

Seems to be a point i just ment that the kernel could be used for other purposes than just be a mimicry of windows , its a completly developed kernel like FreeBSD which you dont come that often across. Why nearly no one use BSD or BEos (and so on) kernels as kernel seems to be also a mystery(only know one Debian distribution with BSD kernel).

Why nearly no one use BSD ... kernels as kernel seems to be also a mystery

Seriously?

You're not looking hard enough; there are loads of machines out there running flavours of BSD.

dstyl wrote:-Not Linux(Everything uses Linux monocultur some variability would be nice in my opinion)

I don't see how that is an advantage. As much as variety is nice when it comes to operating systems, having a lot of different projects all using the same kernel ensures greater compatibility. Seriously, we're technologically advanced enough now to understand why having multiple completely independent platforms is a bad idea.

Hi,

onlyonemac wrote:

dstyl wrote:-Not Linux(Everything uses Linux monocultur some variability would be nice in my opinion)

I don't see how that is an advantage. As much as variety is nice when it comes to operating systems, having a lot of different projects all using the same kernel ensures greater compatibility. Seriously, we're technologically advanced enough now to understand why having multiple completely independent platforms is a bad idea.

We're technologically advanced enough to know that something designed specifically for one purpose (e.g. general purpose high throughput servers) is never ideal for a very different purpose (e.g. real-time systems, fault tolerance, low-latency desktop, low-latency networking/server, small embedded systems, etc).


Cheers,

Brendan

In my opinion, ReactOS is a general-purpose Operating System, like Linux, so onlyonemac's point is relevant to the current discussion.

Embedded systems, handhelds, real-time systems, etc. all have their particular requirements; that is irrelevant to a discussion about ReactOS.

Assuming that the ReactOS kernel is fairly similar to the actual NT kernel, the kernel itself is pretty much agnostic to the userspace API.

NT was designed to support different subsystems/"personalities" and, in the earliest stages of development, it wasn't even going to be used for "Windows" at all (it started life as a next-generation, portable kernel for OS/2). Previous releases supported OS/2 (16-bit) and POSIX subsystems, in addition to the Win32/64+WOW subsystem that's always been "dominant".

I'm sure that, with enough time and expertise, you could take the ReactOS kernel and build an entirely new userspace API, resulting in an OS that looks/feels nothing like Windows. If that's what you want to do, go for it.

As for legal issues, ReactOS is a clean-room reimplementation, however, that doesn't necessarily solve all possible issues. While it's obviously not directly infringing copyright, there are still possible (and quite likely) other issues, especially with patents. I'm sure that if Microsoft wished to do so, they could at the very least find enough "uncertainty" to destroy the project. The fact that they haven't is probably because they don't see it as a threat.

It should be obvious that Windows 7 is "frozen", Microsoft won't be making any further changes to it (except for bug fixes).

Hopefully the ReactOS team can get their implementation 100% compatible with Windows 7 64bit version.
Virtually all devices today have Windows 7 drivers and it would be an excellent achievement if ReactOS can use any Windows 7 driver.

Reactos started 17 years ago, and is still alpha software. I suspect that by the time it is 100 % Windows 7 compatible (even supposing that would ever happen) Windows 7 will have as much relevance as Windows 3.1 does today.

Well yeah, that's the main issue behind ReactOS. As much as it is an incredible achievement, it lags behind Windows severely, after many years of development it's still not on level of NT 5.X, which is officially an obsolete platform these day.

Brendan wrote:We're technologically advanced enough to know that something designed specifically for one purpose (e.g. general purpose high throughput servers) is never ideal for a very different purpose (e.g. real-time systems, fault tolerance, low-latency desktop, low-latency networking/server, small embedded systems, etc).

...which is why Linux can be optimised for many different applications. Saying that Linux is "designed for high throughput servers" is a severe overstatement. I can only conclude that you are unfamiliar with Linux.

Brendan never said that Linux was designed with that purpose in view. I don't think that he was making any points about Linux per se but was commenting on the rather naive suggestion that one operating system can cover all uses. I agree with him that different applications can call for different operating systems.