I got this "idea" stuck in my head for "ages". And I think that now (with KDE4.2 out) is the time to post it.
But before that let's take a look at what I think (not a KDE developer or anyhow tied to KDE, I am just a normal user) will be the future of KDE.
So now with KDE4.2 out which is first really stable / feature complete whatever you want to call it KDE4. KDe4 has a base on which to build on. From now on it's "just" tie the backends and frontends, porting apps to KDE4 and plasma (that why plasma is now plasma-desktop, so there is no confusion with apps that will use plasma /take a look at new Amarok to see what I'm talking about...).
Given the speed of things I am definite that KDE can make all things happen to the KDE4.5 release.
Then all KDE need is multi-touch support and that's that.
So completely theoretically speaking after KDE4.5 (or which ever KDE will have all of the above) there is no need for new releases.
From then on is just adding new plasmoids for apps, plasma desktop...

So theoretically (if I'm correct) devs will have more time .

Therefore I would really like to see a KDE kernel that would be in pair with flexibility to KDE. Something like plasma just made for kernel.
Now I'm no computer scientist, I'm not even anything near; I'm actually political scientist; so I might be talking stupid stuff right now but I just have to get this idea out of my head).

Even though that monolithic kernels rock and micro kernels are harder to code (I've skimmed through the whole Tannenbaum vs. Torvalds debate) but I feel that a more modular (therefore micro kernel) would be more suited and or "hybrid" kernel (but I'm not sure about it).
Anyway I know KDE is DE and additional software and not an OS and probably even not trying to be. But it would still be cool to have another kernel on KDE list :shade: ...
Reasons why:

• Independence
• Devs know what exactly is in it
• It can have better multi-touch support...
• Built on something other than X and therefore having Kwin effects on all kernels, as it would be tied to KDE...

Don't forget this is just an idea. I'm totally pleased with Linux or any other X based kernel and even Windows. I just thought it might be intersting to share this idea with everyone.


[size=xx-small]Also this has nothing to do with Linus jumping ship and changing to Gnome. Honestly[/size]

its great to run on a predefined hardware etc

for example itd be great for open pda / pokedex / phones etc

for compatibility with a wide range of hardware its better to build for a variety of kernels - what kde is doing now




there are still enough things to fix and make

a lightweight version of kde (with resources use like kde 3.5) for better performance (kde 4 is laggy a bit on most computers and unusable at all on computers below pentium 4)

versions of kde for portable devices with existing kernels (linux etc)

improve dbus or trash it and make something of kdes own

add config options to apps (up to the amount there was in kde 3 and more !)

a powerfull synthesizer pack (what arts was aimed to be in kde 3)

computer games and game engines




now you say thats enogh. tomorrow you won't believe it. bill gates once said a pc will never have to have > 640 K ram

The Gnu group has been working on their own micro kernel for almost 20 years and it still is not in a usable state. Check out information on Gnu Hurd. Making an entire new kernel from scratch is a massive and difficult job in and of itself, but further getting it supported by application developers and device driver manufacturers would be an inordinately difficult task. Linux has been around since 1992 and only within the last year or so have the last serious hold-outs amongst device manufacturers finally started to release device drivers for Linux. Linux is still treated like a second-class citizen by many companies that do support it. And add to that the fact that, as an open-source project, they would need a lot of people working on it, people they would probably have to still from the development of other, more established kernels like Linux. I would rather KDE devs focus their attention on perfecting KDE, not get distracted by a foray into completely different territory and not steal attention away from the really good kernel we already have (linux) but that can always be improved as long as there are enough people willing to work on it. If you look at the KDE project recently, rather than trying to narrow its focus by only working on one OS it is now trying to support every major OS (including windows, mac, and the two unixes).

There is always major work to be done under the hood in KDE. For instance, your example of multi-touch. Multi-pointer support is being implemented in X if it isn't already finished by now. Multi-touch would just be a special case of that (you could accomplish the same thing with two mice, or 10 if you're a squid). However, I do not think KDE in general or kwin in particular is set up to support multiple pointers. This is not something that has to be changed in the underlying architecture, it is something that is there but has to be supported in KDE. Compiz already supports it. There are a lot of such underlying technologies that are not supported in KDE. There are some that are really only supported in KDE, like ARGB visuals, but there is still much more to do. And new capabilities and technologies are always coming in to play.

I got confused by the time I got to the fifth line or something in your post as I think you have completely misunderstood how a Linux-based computer works.

Warning: I have greatly generalized and simplified the following information. It's not politically correct but if I go into any further detail there's a high chance nobody here will understand it. ^_^;

The kernel is the absolutely lowest level of the software stack, it's the thing that sends the binary signals to the hardware.
KDE is a set of applications and libraries and is at the absolute highest level of the stack, it never sees the binary signals and it doesn't care what kind of hardware it is on.
In between them is X, the piece of software that takes the graphics from the applications and sends them to the kernel for displaying.

Every single one of those is a completely different project that require completely different programming knowledge.

An application developer doesn't need to know how a computer works at all, everything they need is handled by the libraries.
An X developer requires decent computer engineering knowledge and mathematics as their software deals mostly with moving memory from one internal device to another and transforming said data into different formats efficiently.
A kernel developer requires advanced computer engineering knowledge as they are now dealing with electronic signals. X drivers are usually also kernel modules and can also come under this category.

The worlds best KDE developer probably can't even write an X extension, don't even think about a kernel.

--

Oh and KDE 4.5 will nowhere be complete. It took all of KDE 2 and KDE 3 to get all the features found in KDE 3.5.10.

--

... And most of what I think you are trying to explain just requires some extra X extensions, not really that difficult but trying to find someone willing to work on X will be the problem.

@ash
As you could see in how-lightweight-is ... 28523.html it is possible to have a functional KDE4 desktop on much less powerful hardware than P4.

BTW, what's wrong with D-Bus?

@Zarin
You're not really on the right track with the claim "A kernel developer requires advanced computer engineering knowledge as they are now dealing with electronic signals.". Everything that comes to the kernel is already in a digitized form. For example the power fluctuation that occurs when you press a key on your keyboard (obviously, it doesn't switch instantly from x volts to y in a micro second - it has a curve which stabilizes on y) is not handled by the kernel, but by the keyboard controller.

But your point is on the dot (so I have no idea why I wrote the last part at all )

@Primoz
With that said, if KDE developers were good at writing kernels, and most importantly if they liked it (everything can be learnt, but affinity is not something that is not easily changed) they would all work on linux (or some other existing kernel). Which would be pitty, since KDE would not exist

All have valid points. I really didn't expect anything else. And I do agree that there are kernels that do great job and the writing of a new kernel is a lengthy project (like the HURD).
Can anyone here explain what Solid is trying to achieve.

And when do you think that KDE will be "feature" complete so that from then on
there would be only matter of developing new widgets, an applets.
5.5?
How long do you think that plasma will stay the part of desktop.
May be it's a bit conservative of e, but I think this is it. Given that everything is plasmoid (even panel) there is just a matter of developing plasmoids to replace any part of a plasma...

Careful! You never know! I'm pretty sure that at one point people were convinced that two computers would supply all the computational needs of the U.S. (can't find the quote) They obviously didn't predict all the uses we have for computers now.

The problem is that code gets bloated, new ideas for structuring code are created, and new ways of interacting with a computer are constantly being created, and any library gets stretched and stretched until it reaches a point where it really just can't bend any more. Ideally, yes, plasma could be malleable into any requirement, but realistically, all that flexibility is just an abstraction of a finite set of inflexible parts that were designed for a specific purpose.

The same with the rest of KDE. Eventually, continually adding new ways of interacting with hardware with solid, or accessing sound with phonon, or storing data with akonadi, or making metadata with Nepomuk.... will make the code start to get more and more bulky, and more and more inefficient for the tasks it is being applied to, to the point where it becomes difficult to maintain and extend. (this is why KDE4 was created in the first place)

Because of this, we can never really reach that Shangri-La were we no longer need to improve the software.

And thats why for as long as we have computers, we are going to need programmers