Why not consider a hybrid os?

Here you can discuss ReactOS related topics.

Moderator: Moderator Team

Z98
Release Engineer
Posts: 3379
Joined: Tue May 02, 2006 8:16 pm
Contact:

Re: Why not consider a hybrid os?

Post by Z98 »

Creating an OS that can use both Windows and Linux drivers is more complex than creating one that is designed to work with just one of the two.
fred02
Posts: 551
Joined: Thu Nov 22, 2007 5:54 pm

Re: Why not consider a hybrid os?

Post by fred02 »

PurpleGurl wrote:
zed260 wrote:well strictly speaking i think a much more viable option short term option is a duel kernal letting you run ether windows or linux drivers (but not linux apps)

that would make a lot more sense anyway of course later on one could depreciate the linux part of the code and eventualy elimnate it in favor of windows drivers...
Now that could prove interesting, and whoever writes it could call it Chimera or something. It could use a modified Linux kernel from the most compatible Linux flavor with Linux drivers, but expose a full set of NT APIs and be fully compatible with Windows software.
A problem I see with these setups is that Linux does not have a stable API for it's own drivers. So for such an hibrid OS one will have to also (sup)port synchronisation with the regular kernel, in order to stay current with the latest drivers.
mrugiero
Posts: 482
Joined: Sun Feb 14, 2010 9:12 am

Re: Why not consider a hybrid os?

Post by mrugiero »

fred02 wrote:
PurpleGurl wrote:
zed260 wrote:well strictly speaking i think a much more viable option short term option is a duel kernal letting you run ether windows or linux drivers (but not linux apps)

that would make a lot more sense anyway of course later on one could depreciate the linux part of the code and eventualy elimnate it in favor of windows drivers...
Now that could prove interesting, and whoever writes it could call it Chimera or something. It could use a modified Linux kernel from the most compatible Linux flavor with Linux drivers, but expose a full set of NT APIs and be fully compatible with Windows software.
A problem I see with these setups is that Linux does not have a stable API for it's own drivers. So for such an hibrid OS one will have to also (sup)port synchronisation with the regular kernel, in order to stay current with the latest drivers.
IIRC, that statement only applies to staging drivers. There are several API's for drivers, depending on the device it's for, and they're stable enough to let Ubuntu (and I guess other distros) backport drivers from newer releases to a given stable version.
Z98
Release Engineer
Posts: 3379
Joined: Tue May 02, 2006 8:16 pm
Contact:

Re: Why not consider a hybrid os?

Post by Z98 »

Source compatible does not equate binary compatible.
zefklop
Developer
Posts: 114
Joined: Sat Feb 11, 2006 8:47 pm

Re: Why not consider a hybrid os?

Post by zefklop »

PurpleGurl wrote:
zed260 wrote:well strictly speaking i think a much more viable option short term option is a duel kernal letting you run ether windows or linux drivers (but not linux apps)

that would make a lot more sense anyway of course later on one could depreciate the linux part of the code and eventualy elimnate it in favor of windows drivers...
Now that could prove interesting, and whoever writes it could call it Chimera or something. It could use a modified Linux kernel from the most compatible Linux flavor with Linux drivers, but expose a full set of NT APIs and be fully compatible with Windows software. Similar to Linux+Wine, but tightly integrated, inseparable, and fully designed to run only Windows applications. That could be more FOSS friendly since most of the available drivers would be open source.

The above would be outside of our goals, but could be another viable alternative. Once we get to a good finishing point, other GPL projects are welcome to use our code and mix it up with Linux any way they see fit.
I think that WINE considered writing ntoskrnl as a linux kernel module. Or they might do so in the future.
mrugiero
Posts: 482
Joined: Sun Feb 14, 2010 9:12 am

Re: Why not consider a hybrid os?

Post by mrugiero »

Z98 wrote:Source compatible does not equate binary compatible.
Of course it doesn't, but he was talking about Linux drivers, which usually you have got the source (and in the case of binary drivers, most vendors updates it in a regular basis), so the source compatibility (for Linux drivers only) is just enough, I think.
Anyway, I don't really see the point of making an hybrid OS, not at least before getting the specific OS for Windows binary compatibility done.
It's way more code to maintain, which means less actual work invested per dev in the development of the NT side; I mean, if you have to solve bugs caused by the "merging" of the architectures about every commit, you end up losing time for figuring out how Windows work and implementing it.
Order24
Posts: 3
Joined: Thu Dec 08, 2011 9:49 am

Re: Why not consider a hybrid os?

Post by Order24 »

instead of a hybrid OS why not take windows singularity as a base and build a new OS all together. Also for windows singularity it was released for academic non-commercial so you can use it for said purpose.
PurpleGurl
Posts: 1790
Joined: Fri Aug 07, 2009 5:11 am
Location: USA

Re: Why not consider a hybrid os?

Post by PurpleGurl »

Order24 wrote:instead of a hybrid OS why not take windows singularity as a base and build a new OS all together. Also for windows singularity it was released for academic non-commercial so you can use it for said purpose.
That opens a new possibility. Have an OS that is natively half-way between the two, or an OS that could lend itself to extension interfaces that could run nearly any app (even MS-DOS and CPM if anyone is so inclined to write front-ends for those). I guess that is sort of what was tried with the POSIX approach. A kernel back-end which allows for writing many OS front-ends could be nice, though I think programs are best run when the OS is written from the ground up with that software in mind. You can run many types of programs in the same OS, but with the cost of more bugs, more debugging time, possible mediocre performance, and more space used.

Anyway, what we need to do is just worry about Reactos as it is now. When when reach our initial goals of a Windows 2003 compatible OS with a stable XP-like GUI are reached, other FOSS teams are more than welcome to use the code as they wish within the scope of the GPL. So I guess those who dream of hybrids and stuff should hold onto them until a mature GPL Windows kernel and API set is achieved. Or they could form their own project and use what is available now and build from there. Of course, many proposing such should start by learning Windows' internals first and help out here if possible. Then if they are still interested in a hybrid, they could go on and learn Linux internals if they don't know about them already, and then go from there.
Radhad
Posts: 605
Joined: Wed Apr 12, 2006 5:09 pm
Contact:

Re: Why not consider a hybrid os?

Post by Radhad »

Order24 wrote:instead of a hybrid OS why not take windows singularity as a base and build a new OS all together. Also for windows singularity it was released for academic non-commercial so you can use it for said purpose.
Singularity (that's the real name) does not run any application written for Microsoft Windows NT-like systems. It is just for researching new technologies and methodologies in operating systems. For example, it has a completely different kernel than NT or Linux.
Order24
Posts: 3
Joined: Thu Dec 08, 2011 9:49 am

Re: Why not consider a hybrid os?

Post by Order24 »

singularity is an OS that was developed by Microsoft research so calling it Microsoft singularity is also correct and I'm also well aware of the fact that it has a completely new kernel that isn't based on 40+ year old tech like OS X, Linux, BSD and Windows NT. the singularity kernel was built on safe languages in its very design its completely safe from viruses, malware and the like its also very extensible so one day i believe it will be able to load software designed for windows it's only a matter of figuring out how. the 40 yr old OS that all the operating systems can be traced back to is called Multics that's why i said use singularity as a base. to create a better operating system with a new GUI designed for it would make for a kick ass OS. also from what i read its Fairly easy to implement new drivers to extend hardware compatibility. current OS development requires creating a bunch of bugs in the system to test it then figuring out how to fix them singularity was developed from the ground up to create a system so that stage of the development process would become completely unnecessary in the future there by saving development time and cost while still delivering a safe and secure operating system.
PurpleGurl
Posts: 1790
Joined: Fri Aug 07, 2009 5:11 am
Location: USA

Re: Why not consider a hybrid os?

Post by PurpleGurl »

Order24 wrote:singularity is an OS that was developed by Microsoft research so calling it Microsoft singularity is also correct and I'm also well aware of the fact that it has a completely new kernel that isn't based on 40+ year old tech like OS X, Linux, BSD and Windows NT. the singularity kernel was built on safe languages in its very design its completely safe from viruses, malware and the like its also very extensible so one day i believe it will be able to load software designed for windows it's only a matter of figuring out how. the 40 yr old OS that all the operating systems can be traced back to is called Multics that's why i said use singularity as a base. to create a better operating system with a new GUI designed for it would make for a kick ass OS. also from what i read its Fairly easy to implement new drivers to extend hardware compatibility. current OS development requires creating a bunch of bugs in the system to test it then figuring out how to fix them singularity was developed from the ground up to create a system so that stage of the development process would become completely unnecessary in the future there by saving development time and cost while still delivering a safe and secure operating system.
Interesting, but nearly opposite from our stated goals. Singularity is based on managed code and on technology that is probably patent encumbered. Our goal is compatibility with what already exists, using existing 3rd party drivers, and keeping a familiar GUI. Plus it is written in a language which is incompatible with Reactos.

I'd still be using Windows 98 and IE 6 if they were compatible with what exists today, or at least 2000, since NT-based has always been more stable than 9x-based. That was the look and feel that most resonated with who I was, and the industry who thinks it is all-knowing and knows more than the users took that all away from me. I found a nice skin to give the IE6 interface to Firefox, but it wasn't quite the same, and then Firefox upgraded and the add-on didn't. The current trend of splitting browsers into many tasks, while supposedly more stable, has also increased what I believe is bloat (all the major contenders do it this way now). Likewise, I'd rather use the legacy GUI. When I went from 95 to 98 and later 2000, and then XP, I stripped out and disabled things back to Windows 95. For me, performance is what I want to see, and aesthetics is very low on the list. When I upgrade it is nearly always for compatibility or bug fixes, never looks, and often I don't like the looks of the new stuff since the industry is becoming more and more hostile to people similar to the "geeks" who started it all and those who want to control every aspect of what is going in in their PC.
mrugiero
Posts: 482
Joined: Sun Feb 14, 2010 9:12 am

Re: Why not consider a hybrid os?

Post by mrugiero »

Order24 wrote:singularity is an OS that was developed by Microsoft research so calling it Microsoft singularity is also correct
That's right, but you said
Order24 wrote:windows singularity
which is not correct, as it's not based on any Windows technology.
Order24 wrote: and I'm also well aware of the fact that it has a completely new kernel that isn't based on 40+ year old tech like OS X, Linux, BSD and Windows NT. the singularity kernel was built on safe languages in its very design its completely safe from viruses, malware and the like its also very extensible so one day i believe it will be able to load software designed for windows it's only a matter of figuring out how. the 40 yr old OS that all the operating systems can be traced back to is called Multics that's why i said use singularity as a base. to create a better operating system with a new GUI designed for it would make for a kick ass OS. also from what i read its Fairly easy to implement new drivers to extend hardware compatibility. current OS development requires creating a bunch of bugs in the system to test it then figuring out how to fix them singularity was developed from the ground up to create a system so that stage of the development process would become completely unnecessary in the future there by saving development time and cost while still delivering a safe and secure operating system.
Windows NT is not based in anything close to multics. In fact, the technology used (an hybrid kernel, with features of both monolithic and microkernel designs) wasn't even in the paradigms of that time IIRC.
Also, the idea behind Singularity (a managed code based OS) is not that original, as you can see here and here (sorry for needing two links, but as far as the specific term is coined by MS, it's not mentioned in the first article).

Implementing drivers would be probably equally hard with that, because you keep needing to deal with low level.
PurpleGurl wrote: Interesting, but nearly opposite from our stated goals. Singularity is based on managed code and on technology that is probably patent encumbered. Our goal is compatibility with what already exists, using existing 3rd party drivers, and keeping a familiar GUI. Plus it is written in a language which is incompatible with Reactos.

I'd still be using Windows 98 and IE 6 if they were compatible with what exists today, or at least 2000, since NT-based has always been more stable than 9x-based. That was the look and feel that most resonated with who I was, and the industry who thinks it is all-knowing and knows more than the users took that all away from me. I found a nice skin to give the IE6 interface to Firefox, but it wasn't quite the same, and then Firefox upgraded and the add-on didn't. The current trend of splitting browsers into many tasks, while supposedly more stable, has also increased what I believe is bloat (all the major contenders do it this way now). Likewise, I'd rather use the legacy GUI. When I went from 95 to 98 and later 2000, and then XP, I stripped out and disabled things back to Windows 95. For me, performance is what I want to see, and aesthetics is very low on the list. When I upgrade it is nearly always for compatibility or bug fixes, never looks, and often I don't like the looks of the new stuff since the industry is becoming more and more hostile to people similar to the "geeks" who started it all and those who want to control every aspect of what is going in in their PC.
I usually do the same. Anyway, in Linux I always try to get a nice look while keeping it fast, but at most, I use some transparency or shadows, nothing too consuming. I don't think Singularity will be good for performance, running under a VM.
Order24
Posts: 3
Joined: Thu Dec 08, 2011 9:49 am

Re: Why not consider a hybrid os?

Post by Order24 »

as long as the kernel still has monolithic designs it's still based on the same development ideology as Multics, the devs that worked on the NT kernel and singularity have stated this when talking about the reason for starting the singularity project to begin with this is the link http://research.microsoft.com/en-us/new ... arity.aspx and the problem with inferno is that all programing must be done in the limbo language where as singularity use's C# and as far as drivers go http://singularity.codeplex.com/ describes how there made and they have used elements of singularity to make win 7 & win 8 more secure and if singularity won't work as a base the they have a new spin off called verve http://news.softpedia.com/news/Download ... 1975.shtml but while these research projects are not based on windows technology some windows technologies are based on the research projects
Post Reply

Who is online

Users browsing this forum: No registered users and 45 guests