Wayland does indeed follow the UNIX philosophy.

Oddly XOrg and Wayland aren't the only currently worked on rendering agents. Doesn't DirectFB work similar to X?

DirectFB is more similar to wayland in that they both run on top of the kernel framebuffer.
It is also older then wayland and supports input.

I remember recompiling SDL with directFB support so i could play vulture-nethack without X.
It ran software only on top of nothing but the kernel, so i thought it was cool. :)
Had some input problems thou, don't know if it got better.

So it's already a trainwreck. Congratulations!

I didn't say that anywhere.

I see this as the final phase of the downfall of the "free desktop". I don't think anything beside Android's and Apple's display manager on top of Linux/Unix will survive in the long term. Not because these concepts are so great, but because they are there, proven and they work. The latter is also the single reason, why X11 is still around and why I will stay with it for the time being.

I don't think there is really a place for two or more additional incompatible implementations created from scratch. That's not how software ecosystems work.

It was already a trainwreck when those people started to maintain it. WHat the hell didi a print server in the print server? Nonetheless those people did their best to fix what could be fixed, but at one point they had to acknowledge that certain things can't be fixed, because they were inherent to the X11 design. This was the point were they decided that it is better to start over with a clean slate instead. And they did it in a way that they would come up with the best possible solution they could think of. Why do you think that they alone needed 5 years to get to version 1.0 of the protocol definition?

Sorry, you didn't say it is against the Unix philosophy, you did say it is against proven Unix principles, in post #148. I would like to see an explanation for that.

Obviously, companies like Valve, Intel and Red Hat see that different. We will see who is right about that.

It's not so much even X11 really. Most stuff is starting to be handled by OpenGL based libraries almost to the point the standard XVideo display is being deprecated for things like GlamorEGL that renders 2D drawing in hardware as opposed to MesaLib which renders 3D in hardware via the DRI.

Fairly soon, you won't have to worry about it on Wayland as fairly much anything and everything could be passed through OpenGL/MesaLib in some way, shape, or form to have one drawing matrix for all video cards with proper supported DRI drivers.

All in all, Wayland actually is a good idea, but it's not exactly ready yet. When it's completed, you won't even really notice the difference really.

Isn't xwayland going to take care of x11 compatiblity?

http://wayland.freedesktop.org/xserver.html

XWayland land will be merely a legacy library for applications not updated for the Wayland API. Most applications that utilize the current x11 library will drop in with the xwayland library.

When completed, you won't see a difference. Each desktop environment will look exactly the same, work the same, applications will work the same, etc.

The only difference you'll see is the Compositor, remote access, and other features will work differently in the aspect, they'll work like their supposed to and not have issues with other parts of the x-server.

So as long as there are no stability issues the move to Wayland shouldn't be an issue. Since there are x.org developers involved I don't expect any compatibility issues. After being tested I would favor rolling over to it.

The only issue is you might see a lot of older video hardware without proper drivers get pushed into support through VESA and the MesaLib llvmpipe driver. At best AMD, Nvidia, and Intel will probably get full DRI2 support and drivers.

Of course, for approximately 18-24 months or two Slackware releases. After that I bet it will be abandoned and left unmaintained.

Further reading: http://tirania.org/blog/archive/2012/Aug-29.html

Anyway: Debian/Ubuntu's init decision downgraded the Linux desktop officially to a legacy platform for me. I'm now looking into alternatives and prepare for the phase-out of all my Linux desktops. I want to depart early before the fallout comes down.

So I basically don't care about Wayland anymore.

What does the init system have to do with wayland? As far as being an end-user the init system & Windowing systems shouldn't matter as long as they Just work. Its more of a headache for system admins & developers.

Did you realize the blog you referenced was from 2012?

Seems like the Linux Desktop should have been dead already. Linux as desktop will probably be for a niche group which it has been since the start. Chromebooks & Android will probably be the dominant flavor of linux for the masses anyway once everyone uses the "cloud" for everything.


But as far as Wayland I am sure it will be supported as long it used in most of the major Linux Distros.

With regard to this being a Wayland thread, I agree init references seem out of place. But both are big changes being more or less forced on the rest of us, and they have far reaching consequences for us beyond the mythical "end user" use case, so such sentiment is often seen these days.

With regard to the second part... that attitude is quite literally the death of free software... you should reconsider it.

The argueability of whether or not an init system will be the downfall of Linux is debatable due to the level of chaos consuming everything, but Wayland is actually good for UNIX on the whole. BSD, Illimos, Solaris, HP-UX, and every other X using solution are all getting Wayland ready for implementation.

Wayland doesn't have to require any specific init.

Wayland is simply a modernized X-Server with a built-in Compositor and modernized Remote Access Networking layer with its rendering layer based in EGL and XVideo extensions for maximum levels of hardware acceleration and API extension support.

As far as getting ready to abandon GNU/Linux on the whole, well yes, start learning now and contribute to distributions you feel are worthy successors, but equally keep supporting Slackware, Gentoo, LFS, and other distributions that promote choice and freedom within the system.

It's either that or get those all those forced into deprecation projects back into solid developments, and push the crap out of OpenRC and fight back.

To be honest, I had the init system more in mind myself.

I am deeply troubled by the lack of knowledge or care for the FREEDOM aspect of free software and the idea that, "Hey if it works, it's cool!". The abandonment of the huge value of the Unix ideas for "end user" trinkets is also very distrubing. The two opinions cannot coexist in the long run.

I have not decided myself to abandon GNU/Linux, but I fear for its future enough to hedge all my bets at this time.

If I found that I no longer wanted to go down the path, such as due to systemd, I would entrench with the last non-systemd Slackware and go from there... I have dipped into the BSDs but not to any depth, still an option.

But back on topic, I am not particularly knowledgable about Wayland, except that it is not client/server and therefor not usable across a network. I do occasionally export an X session and would consider loss of that ability to be... a loss. I always hate to lose ground...

I follow the Wayland threads here and will form an opinion as we go forward.

I fully believe in Slackware and Linux (kernel) itself. I'm quite sure, that they won't screw up.

But that isn't just enough for a working desktop platform. A big part of the userland upstream is now axing the unix legacy (basically all the stuff that still works reliably and has stable interfaces) for creating a new "free desktop operating system". And I'm just not interested in this OS at all. Especially not if it is introduced into all Linux distribution through the backdoor.

I think the desktop developers believe that their product is so great, that people accept the "complicated" Linux beneath it just to get access to their great invention. That's why they try redesigning the base without understanding it. My reality is that I accept some more or less mediocre GUI stuff, because it runs on top of a stable and powerful Unix. I sometimes use GIMP, because it's free and runs on Slackware, not because it is the best image editor on earth. And I have non-free third-party applications from ISVs which need stable interfaces to continue to work. That is what I'm interested in.

When that stable and powerful Unix base is gone, there is not a single reason for me to run Wayland, Xwhatever, KDE or anything else on top of a new Fedora-LennarX. (Just because something is free doesn't make it interesting to me.) As soon as my desktop gets turned into something like that, I will move on to other pastures.

Here are a few proposals:
Akaros integrates code from Plan9, now dual licensed.
Maybe use OS/2 as a basis. Browser qupzilla is available for it.
Oh and eComStation claim to be fully compatible with OS/2 (non free).
Not to forget Haiku, for personal use. Qupzilla is also available for it.

PS Just trying to fight insomnia ;)

eComStation is fairly nice for an OS. I had a trial of it a while back for an old machine I was using and it worked really well. It's kinda iffy on newer hardware so unless you want problems you might want to stick to hardware from the start of the Windows XP era.

Another odd-ball OS out there is SkyOS. http://www.skyos.org/ I don't know much about it, but it seems to be either some kind of UNIX-like or another independent designed OS, similar to Haiku/BeOS. Looks promising, but progress on it seems to be about the same pace as ReactOS.

Thanks for the recommendations, but I think like Miguel de Icaza and some BSD developers I will most likely end up with a certified proprietary Unix, which has a fair amount of useful desktop applications from ISVs available. That doesn't mean, that I completely abandon Linux.

eCS with its 286 optimized 16 bit kernel and a 32 bit hybrid userland on top of it is something that just doesn't work on post-2010 64 bit PC hardware. OS/2 really is a thing from the past.

Sorry but that's not exactly right man. EComStation is a true 32-bit OS, not a hybrid 16-bit/32-bit mess like Windows 9x was. The original versions of OS/2 were 16-bit, but IBM made a clean transition to 32-bit unlike Windows did with 9x from 3.x which utilized a 32-bit user land on top of a DOS core and 16-bit kernel with 32-bit memory addressing extensions.

OS/2 never was a true 32 bit OS like NT. eCS kernel and file system drivers are 16 bit for compatibility reasons. Also there is a lot of 16 bit x86 assembler code everywhere, which no one can touch without breaking everything. This design (dating back to the IBM AT) is the main reason, why OS/2 never survived the introduction of x86-64.

http://blog.martin-graesslin.com/blo...-trancparency/
I recommend to also read the re4st of the article, I find it very informative.

sometimes i think Martin Graesslin is thinking the new plasma only (or what's it called)
i think wayland will need compositor support for network transparency, unless couple of them come together and make a standard (for compositors over network)

also i don't see nothing wrong with vnc

X12 if its ever made will have its own network transparency
wont be X11, but you will have X12 anyway

also what does dbus have to do with it

As he states, he thinks that network transparency belongs into the toolkit. Anyways, since he is the main developer for Kwin (and therefore also the main developer for making it a Wayland compositor) I think he knows what has to be done to be network transparent.

As he states, X11 network transparency does not work for applications that rely on DBus interfaces, since DBus itself is not network transparent.

sry didn't have time to read it thoroughly

thing is there is no definition of a "compostior"
since a window is basically a textured quad to a compositor, it can do whatever it wants with it
it can rotate it, change the perspective, wiggle it, add shadows or whatever and even map it to a 3D object
(and the client can treat it as a texture buffer and draw in it using cpu or gpu)

i was thinking like when sending a window event decide
IF CLIENT idk QT5.69 THEN SEND qt5.69 api for it
ELSE IF unknown THEN SEND damage events or full textures

and since dbus is a socket thing, why not just send it over the network
(i know its not internets capable, but routing msgs)

just a couple of thoughts

According to an fvwm developer, fvwm is unlikely to ever be ported to Wayland:

http://www.fvwmforums.org/phpBB3/vie...php?f=6&t=3010

.. so I hope the xcompositor works well.

Between that and the "oh, yeah, someday, eventually" attitude towards network transparency (which I use frequently to work around practical problems), I expect to need to hang on to Xorg for a while before making the switch.

Window Managers based on xlib don't necessarily have to be obsoleted, they just need to evolve to give the added functionality needed to the standard Wayland interfaces, controls, etc.

Isn't twm the most basic wm for x11 getting a Wayland counterpart?