Target Kernel

Here you can discuss ReactOS related topics.

Moderator: Moderator Team

shunesburg
Posts: 149
Joined: Wed Feb 21, 2018 3:46 pm
Location: Somewhere in France

Target Kernel

Post by shunesburg » Fri Aug 31, 2018 11:50 am

Hello,
currently the target is NT5 and there is a upper compatibility layer for NT6+.
But in the future, this target would be change or stay like that ?
I think it will be better to made the reverse, target NT6.2/NT10, but keep a subsystem or compatibility layer for previous kernels and drivers.


shunesburg
Posts: 149
Joined: Wed Feb 21, 2018 3:46 pm
Location: Somewhere in France

Re: Target Kernel

Post by shunesburg » Fri Aug 31, 2018 2:41 pm

dizt3mp3r wrote:
Fri Aug 31, 2018 12:14 pm
Search the forum for NT6 compatibility before posting.
I read the forum very often, I know ReactOS is compatible with NT6 (I said in my previous post : "currently the target is NT5 and there is a upper compatibility layer for NT6+").
But I spoke about the target of the ReactOS is NT5 by default, that create problems for NT6 feature implementations and the modern drivers can't be implement too because it's not in the target kernel.
I know ReactOS is in development and for long time and if the target don't change when the project will be stable, it will be outdated (and it's already true).

karlexceed
Posts: 454
Joined: Thu Jan 10, 2013 6:17 pm
Contact:

Re: Target Kernel

Post by karlexceed » Fri Aug 31, 2018 3:39 pm

https://reactos.org/wiki/Missing_ReactOS_Functionality

ROS will get there eventually. Perhaps later the "default" install of ROS will end up looking/acting like Windows 7, but not any time soon.

The cool thing about ROS is that once the kernel is built and stable, pretty much any other Windows version can be built on top. XP just happens to be the current goal, almost like a proof-of-concept. I can see a day where ROS forks into two/three major distros - one targeting XP, one for 7, and one for 10, where at their core they're all the same, but just have different shells and default exposed APIs.

middings
Posts: 1007
Joined: Tue May 07, 2013 9:18 pm
Location: California, USA

Re: Target Kernel

Post by middings » Sat Sep 01, 2018 2:20 am

Back in 2014 there was much discussion about the ReactOS Project's future course as Microsoft's end of support (EOS) date for Microsoft Windows XP and Server 2003 products approached. The ReactOS developers (devs) kept their own counsel. Their consensus was, apparently, don't panic. The devs kept Server 2003 compatibility as the project's development target.
To make a big rocket that flies, start with a small rocket that flies.
With the advantages of hindsight, I believe that the devs took the best course. Had the devs switched the project's design target to Windows 7 in 2014, progress toward a working OS would have ground to a halt for many years. The project would be trying to make a big rocket that flies out of a smaller rocket that did not fly. We would be having arguments over switching the design target again as we approach the EOS date for Windows 7 and Server 2008.

My personal opinion* is that more important than adopting a leading-edge design target is achieving the minimum viable product (MVP) or compelling product offering (CPO) stage. This is the minimum functional feature set that causes people to adopt ReactOS for productive use. It goes beyond 'friends and family' use. I have no idea what ReactOS's MVP application might turn out to be.

* worth nothing

florian
Posts: 453
Joined: Tue Nov 01, 2005 2:19 am
Location: Germany

Re: Target Kernel

Post by florian » Sat Sep 01, 2018 4:04 pm

Sorry middings, but I like leading-edge design - for instance the "next supercar" Fahrradi Farfalla FFX... :o

dizt3mp3r
Posts: 1438
Joined: Mon Jun 14, 2010 5:54 pm

Re: Target Kernel

Post by dizt3mp3r » Sat Sep 01, 2018 5:11 pm

middings wrote:
Sat Sep 01, 2018 2:20 am
My personal opinion* is that more important than adopting a leading-edge design target is achieving the minimum viable product (MVP) or compelling product offering (CPO) stage. This is the minimum functional feature set that causes people to adopt ReactOS for productive use.
Absolutely correct.

EmuandCo
Developer
Posts: 4314
Joined: Sun Nov 28, 2004 7:52 pm
Location: Germany, Bavaria, Steinfeld
Contact:

Re: Target Kernel

Post by EmuandCo » Sat Sep 01, 2018 5:24 pm

First we need a stable kernel and always trying to add some questionable stuff MS did with any newer Windows does not help us here. As soon as NT5.X is good to go we can discuss about that stuff, so I 100% support middings' post.
Image
ReactOS is still in alpha stage, meaning it is not feature-complete and is recommended only for evaluation and testing purposes.

Ancient
Posts: 81
Joined: Tue Mar 27, 2018 11:32 pm

Re: Target Kernel

Post by Ancient » Mon Sep 03, 2018 8:11 pm

EmuandCo wrote:
Sat Sep 01, 2018 5:24 pm
First we need a stable kernel and always trying to add some questionable stuff MS did with any newer Windows does not help us here. As soon as NT5.X is good to go we can discuss about that stuff, so I 100% support middings' post.
God bless you and go for it. Get a stable Win 32 kernel single core NT 5 or so OS up with Windows GUI asap. Delays hurt any potential for this project. Best of luck.

wdstudios
Posts: 35
Joined: Mon Aug 13, 2018 8:53 pm

Re: Target Kernel

Post by wdstudios » Mon Sep 03, 2018 10:53 pm

One of the advantages of keeping NT 5.x as the target is that, if ReactOS ever sees widespread adoption, people in the FOSS community will create all kinds of drivers, compatibility layers, and other support to get it to work on modern hardware or with modern programs that require Win7+, and those goodies can then be backported to XP.

A lot of the more common responses to this question can basically be summed up as "Remember how Duke Nukem Forever was in development hell for 15 fricking years because the designers kept adding more shit, then scrapping it all and starting over from the beginning? Like that time in June 1998 when they switched from the Quake II engine to Unreal? Do you think that ReactOS would benefit from that kind of approach to development?"

Quim
Posts: 231
Joined: Wed Jul 04, 2018 11:45 pm

Re: Target Kernel

Post by Quim » Tue Sep 11, 2018 1:52 am

Any estimation of what is still missing in ReactOS NT 5.2 kernel (a list or something)?

According to WINE project:
https://source.winehq.org/WineAPI/ntoskrnl.html

Quim
Posts: 231
Joined: Wed Jul 04, 2018 11:45 pm

Re: Target Kernel

Post by Quim » Thu Sep 27, 2018 9:18 am

EmuandCo wrote:
Sat Sep 01, 2018 5:24 pm
First we need a stable kernel....
As soon as NT5.X is good to go....
And how do developers will reliaze that NT 5.2 kernel is complete enough ?
According to some posts, nobody (even kernel developers) knows how much is ReactOS kernel completed.

shunesburg
Posts: 149
Joined: Wed Feb 21, 2018 3:46 pm
Location: Somewhere in France

Re: Target Kernel

Post by shunesburg » Mon Feb 18, 2019 1:29 pm

I rise this thread.
Quim wrote:
Thu Sep 27, 2018 9:18 am
EmuandCo wrote:
Sat Sep 01, 2018 5:24 pm
First we need a stable kernel....
As soon as NT5.X is good to go....
And how do developers will reliaze that NT 5.2 kernel is complete enough ?
According to some posts, nobody (even kernel developers) knows how much is ReactOS kernel completed.
The kernel is already stable, but the problem is the driver compatibility (lot of crash is due to driver issues), and a change of target will resolved it.

middings
Posts: 1007
Joined: Tue May 07, 2013 9:18 pm
Location: California, USA

Re: Target Kernel

Post by middings » Mon Feb 18, 2019 1:45 pm

Quim wrote:
Tue Sep 11, 2018 1:52 am
Any estimation of what is still missing in ReactOS NT 5.2 kernel (a list or something)?
"Search before you ask" remains good practice in a technical forum such as this.
I suggest searching the ReactOS Wiki using the word "missing". A link to the wiki is provided on the ReactOS.org home page.

PurpleGurl
Posts: 1774
Joined: Fri Aug 07, 2009 5:11 am
Location: USA

Re: Target Kernel

Post by PurpleGurl » Mon Feb 18, 2019 3:49 pm

wdstudios wrote:
Mon Sep 03, 2018 10:53 pm
A lot of the more common responses to this question can basically be summed up as "Remember how Duke Nukem Forever was in development hell for 15 fricking years because the designers kept adding more shit, then scrapping it all and starting over from the beginning? Like that time in June 1998 when they switched from the Quake II engine to Unreal? Do you think that ReactOS would benefit from that kind of approach to development?"
You have a point... maybe someone could do a loose fork after we get this one even close to where it needs to be... I mean, sometimes, using software as a guide to rewrite from scratch can result in something even better and more stable. I know that was true in modules I've written. You often start out with shots in the dark and keep experimenting until it works.... then you seek to understand what you wrote, and then that's when the "ah hah!" moment occurs and you realize you can do it simpler and yet better. We may think a problem is more complex than it is, and write with a complex solution in mind, and then realize what it's actually doing, overall. From there, you can rewrite it more efficiently.

However, at this point, we want to stay on the same path that was started on and reach some sort of usable state. Once you get it suitable for daily use, then you may be able to see more clearly to simplify things in places.

shunesburg wrote:
Mon Feb 18, 2019 1:29 pm
The kernel is already stable, but the problem is the driver compatibility (lot of crash is due to driver issues), and a change of target will resolved it.
No it is not, and an unstable and partially implemented kernel is part of what is causing the driver compatibility issues. Changing the target won't magically resolve it... since you have to then finish the older components and then add new parts on top of that, with some changes in the old parts. So stick to a simpler Windows version and build from there, and make it solid.

Post Reply

Who is online

Users browsing this forum: DotBot [Crawler] and 2 guests