New Greentea OS kernel design concepts apply to ROS?

[Quim]

New Greentea OS kernel design concepts could be useful to ReactOS?

Strict distinction between wine components and kernel, so we could make a not-NT simple STABLE ON HARDWARE kernel and incrementally add NT features then

Imagine similar to Linux: all-in-one (with device drivers) kernel, stand alone user-space as in wine, but with potential to support NT drivers in future

This allows us to backport most linux drivers as-is and keep everything in "just works" state, instead of never-ending endevaour of reactos to support only NT (buggy and incomplete)


Currently investigating simulation of Greentea userspace atop of Windows... Basically like "Wine on Windows". To see what kernel subset is critical to apps. Maybe to be released as stand alone project too

This is not NT OS, anymore


Some parts of Wine would be in use, of course (because there is no other way to implements some things exctept in a buggy C++ code)


New kernel is Tofita, x64, monolithic (like Linux)

With sandboxes


Compatibility as much as possible, but everything that is too hard to implement is hidden in Tofita


For example ReactOS tries to support say USB driver, while all most every other subsytem is buggy and incomplete, so USB doesn't works anyway

We may just implement USB in a non-NT way, So it "just works" .


#TofitaKernel

Tofita kernel boots into UEFI x86-64, initializes true color graphics, COM and PS/2 keyboard.

Next: initialize PS/2 mouse, load filesystem image into RAM, add simple memory allocator.

#TofitaSimulator

It's time to start load some exe's and hook into kernel32 (currently only DLLs supported).

Simulator aims to simplify development and testing of userspace, without use of virtual machine. It emulates pretty much everything a virtual machine does, except for hardware, but in a fast and easy to debug way. It would be fun to see Greentea-as-an-app.

Yes, with Tofita there will be only LiveCD with full user interface and installer, just like any Linux does Greentea OS kernel design concepts could be useful to ReactOS?

Strict distinction between wine components and kernel, so we could make a not-NT simple STABLE ON HARDWARE kernel and incrementally add NT features then

Imagine similar to Linux: all-in-one (with device drivers) kernel, stand alone user-space as in wine, but with potential to support NT drivers in future

This allows us to backport most linux drivers as-is and keep everything in "just works" state, instead of never-ending endevaour of reactos to support only NT (buggy and incomplete)


Currently investigating simulation of Greentea userspace atop of Windows... Basically like "Wine on Windows". To see what kernel subset is critical to apps. Maybe to be released as stand alone project too

This is not NT OS, anymore


Some parts of Wine would be in use, of course (because there is no other way to implements some things exctept in a buggy C++ code)


New kernel is Tofita, x64, monolithic (like Linux)

With sandboxes


Compatibility as much as possible, but everything that is too hard to implement is hidden in Tofita


For example ReactOS tries to support say USB driver, while all most every other subsytem is buggy and incomplete, so USB doesn't works anyway

We may just implement USB in a non-NT way, So it "just works" .


#TofitaKernel

Tofita kernel boots into UEFI x86-64, initializes true color graphics, COM and PS/2 keyboard.

Next: initialize PS/2 mouse, load filesystem image into RAM, add simple memory allocator.

#TofitaSimulator

It's time to start load some exe's and hook into kernel32 (currently only DLLs supported).

Simulator aims to simplify development and testing of userspace, without use of virtual machine. It emulates pretty much everything a virtual machine does, except for hardware, but in a fast and easy to debug way. It would be fun to see Greentea-as-an-app.

Yes, with Tofita there will be only LiveCD with full user interface and installer, just like any Linux does

Could parts of these concepts be used in ReactOS?

[ThFabba]

This is a bunch of buzzwords thrown together by somebody who has no idea what they're talking about. No, it's not useful.

[Quim]

This is a bunch of buzzwords thrown together by somebody who has no idea what they're talking about. No, it's not useful.

I see now.
I thought it was a good idea but,...

[PurpleGurl]

No. Trying to make a non-NT kernel of our own and then incrementally adding support is not useful. The big problem there is that it causes a bigger waste of time. So a non-standard kernel leads to nonstandard things in the middle, drivers not working, etc.

Backporting Linux won't work as we are not Linux. Not only are the formats so different that a rewrite is better than trying to adapt, the underlying philosophies are different. We and they put the graphics functions in different places, and that is not compatible. If you want to work around that, then when not work to rewrite Linux to be able to use Windows video drivers? Linux people will really hate you for that, call you a heretic, etc. Thus suggesting that we do similar could provoke the same sort of visceral reaction.

And some of the original post sounds like a dig. I mean, where is this "buggy C++ code?" We don't currently use it much except where com object programming is used, such as the explorer. It's a nightmare to attempt that in C.

Having layers on top of Windows makes no sense. Just run your stuff in Windows. I cringe enough hearing about running this in VMs, but I understand it for R&D purposes, just not for end users.

So their approach seems to enable the current problems by trying to replace portions with Linux, which means making the rest in more of a Linux fashion. So that pushes it further from the goal of running NT drivers.

What we need is more manpower. More funds would also help. Ideally, we could have the funds to pay out 10 or more bounties at a time, and enough developers to where that would be possible. But we don't.

And their approach doesn't seem like those who want to do heavy lifting. It is like they just want to cobble together other people's work and end up with another Linux-like chimera that very few will be able to get to work.

[EmuandCo]

Oh come on. Didn't we all already tell more than once what useless tar pit this "project" called Greentea OS is? Stealing here and maybe a bit from there too and then claim they have the uber concept anyone using plain english would call a MESS. I will lock any Greentea OS discussions from now on. Cya ^^