Wayland/Weston Gets Forked As Northfield/Norwood

An anonymous reader writes "Weeks after Canonical announced Mir, Wayland's display server protocol and Weston compositor have been forked. A contributor to Wayland found differing views with the project over desktop eye candy and other technical decisions to the X11 successor, which resulted in forming the Northfield and Norwood projects. The developer, Scott Moreau, has been outted from the project but has provided a lengthy explanation why the fork was needed to advance the Linux desktop."

Comments

Survival of the Fittest

[jarich]

I'd never do it myself, but I'm looking forward to seeing which projects survive and how they change the landscape in five years. X11 was difficult to use for years... let's see what a little competition can do for innovation and usability.

Re:Survival of the Fittest

[Anonymous Coward]

The guys developing Wayland are the core developers of X11.

Re:Survival of the Fittest

[SolitaryMan]

The good thing about Open Source is that if two contributors strongly disagree on something, both are given an equal opportunity to prove their point. In the end, society wins.

Re:Survival of the Fittest

[epyT-R]

blasphemy! if you're going to use a game analogy, make sure you use the right game. shao-khan will collect your geek card and then enslave your soul.

Re:Survival of the Fittest

[Anonymous Coward]

No. Instead of working together to resolve their differences and taking the best technical ideas from both groups to enhance the core product, they have a whiny bitch fest, split the development teams, create two products which end up lesser than the original, create an artificial need for more developers, reduce the advancement of the state of the art, and cause people new to the area to spend twice as long researching which products they should use.

In the end, society loses.

Re:Survival of the Fittest

[Anonymous Coward]

wierd since Kristian wrote AIGLX and DRI2 which are X. Scott is that you?

Re:Survival of the Fittest

[jedidiah]

Blob drivers getting you down? Vmware and Nvidia both solved that problem. Redhat and Ubuntu made it even easier.

It's really not the problem some people like to make it out to be.

BLOBS really only become a problem for companies that don't actually want to support Linux. They throw together a kernel specific binary and that's all you hear from them.

With dkms there's really no excuse at all for that anymore.

The kind of vendor that would give you a single kernel BLOB driver is probably a vendor you don't want to use with Windows either. Some stuff is just crap. Some companies are just crap.

You are going to get some crap in a free market regardless of how much you pander to hardware vendors.

More information

[Darxus]

Everybody involved with the wayland project is happy to see weston (the reference display server) get forked to be developed into a more usable desktop environment. That's basically what it's for, and this is far from the first (ubuntu forked it, ADWC was another fork).

This entire article argued he couldn't do what he needs with a plugin alone, which is not relevant to his problem with the wayland community. The problem was his refusal to use the existing mechanism to retain protocol compatibility by copying the existing protocol code into a new extension and modifying in there: http://lists.freedesktop.org/archives/wayland-devel/2013-March/008172.html

In four pages, he didn't address why he didn't feel like doing that.

Re:More information

[lytles]

nothing he said in there was anything worse than what linus posts, or many open source projects. you could have said "scott - i choose not to make the changes that you'd like. you're free to fork things". instead, you're talking out of both sides of your mouth - claiming that you were being accommodating and then stonewalling him, kicking him when he called you on it, and pretending that he's a bad egg for forking things

if you're going to bad-mouth someone for forking, then you're just playing politics

Re:More information

[slack_justyb]

Because that solution just sounds silly.

The whole point of Wayland, and I could be wrong on this one, is to avoid the mistakes of X. From your solution, we just go back to the days of hack around the core.

The core protocol is flawed and the project shouldn't be afraid to make some sort of shift when there is a pretty good reason for that change. If dude was purposing change for just change sake, yeah I would get it. But the protocol doesn't implement basic window management within the core, and makes it insanely difficult at the plugin level. I, for one, think dude has a point.

That is exactly what happened to X. Everyone was afraid of changing the core protocol, afraid that it would break older stuff. Look where that got it. About a bazillion extensions. At some point backwards compatibility breaks the baker and I know that is hearsay in the OSS community. I know there has been a lot of boneheaded change for change sake forks, but I really don't see this as falling into that category.

Re:More information

[Darxus]

Nobody is asking him to keep it out of the core forever. Just making his modifications in an extension so they can be cleanly added to the core (or his extension can become the core), instead of breaking things as he changes it.

Re:More information

[serviscope_minor]

The core protocol is flawed

Bullshit!

The core protocol is very very good.

There are a fewnow obsoloete bits like the bitmap graphics and fonts, but the rest of the core is fundementally sound.

So, unless you can provide some evidence that the protocol is flawed, then go back under your rock.

Oh yeah and Keith Packard: quotes don't count. He might be a core X11 developer but he's also a one man FUD machine, which is really sad.

About a bazillion extensions.

So? Let me repeat that, so?

The protocol is *designed* for extensions. Extensions are simply the X equivalent of adding more API calls.

Look at Linux: it is a POSIX kernel, but with LOADS of extensions! Does anyone whine "oe woe posix is flawed linux has so many extensions it's fundementally broken lets nuke it stard from scratch and by the way no one needs multi user right?"?

Seriously, who the fuck cares if they're "extensions" or "core", any more that if I care if the networking zero copy super low latency fast stuff is an extension or in the core POSIX system call interface? A clue: no one cares.

No one cares on Windows either! Turns out that Windows 8 implements all those Win32 calls right back to the Win32S (a Windows 3.X era thing) API. Do you hear anyone complaining that because Windows has some old API calls in it that it is fundementally flawed and the mere existence of old APIs taints it? No, because that would be stupid.

Likewise: you know OSX used to allow you to run OS9 and earlier binaries! My got it supports old stuff!! It must be fundementally flawed! Oh my god! Every system supports some degree of backwards compatibility!! They are ALL FLAWED! We must rewrite them all in javascript on the cloud in HTML5 except that all the browsers can render HTML 3.1 so they are flawed too!!!

Here's what astonishes me: people hate X so much that they complain (a) when it's too old and (b) when the developers add API calls to make it more up to date!

Actually, I'm in favour of updating the core protocol, or adding some hefty extensions to reduce latency NX style---though XCB actually is a fair improvement over xlib---but Wayhand is not the answer to that because it removes basically everything.

Re:More information

[serviscope_minor]

Optimise the common case' has been wisdom in the *nix world from before we were born. In the case of X, what was the common case when it was designed is not the common case now, and X does not optimise that common case.

X does a fine job of bringing up a bunch of windows, distributing events and providing both direct and indirect hardware access.

Also, optimize the common case is not the same as trash old stuff and remove features for something not demonstrably better.

That common case is local display, using graphics hardware which is built around 3D and OpenGL.

Nope. The common case is still largly a bunch of 2D windows.

I don't do all that much 3D, not that it matters...

What the display subsystem needs to do is to efficiently make available the display hardware capabilities of the machine it is running on, in a way that is easy for people to program.

Certainly.

Then there is the question of what a modern desktop environment needs and how to efficiently deliver that.

I suppose you could say the goal is to build a quality desktop environment, yes.

The design assumptions of X, and the need to work around things using extensions and suchlike, make things harder than they need to be.

Seriously, what is this fetish with extensions that you people have. They are not "workaronuds" they are new API calls. Eveyr system on the planet introduces new API calls when the time comes for it. What is this bizarre double standard holding up X to be something magically pure which isn't allowed new API calls without them being "hacks" or "workarounds".

If you want your 'evidence', take a look at the size of code and execution time required to do basic and complex tasks using X vs similar situations on Mac and Windows.

This is the same X that allows direct rendering, giving programs very direct and efficient access to the hardware and often runs games a few FPS faster than the Windows version of the same game? (Is OSX even in the running here?)

That X? The one that's already better than the two systems you are proposing are better?

Unless you're running a compositor, then X almost completely gets out of the way. If you are compositing, then it almost completely gets out of the way.

For 3D graphics... but a desktop is much more than just 3D graphics output.

Standards

[ArhcAngel]

Standards

"Advance the Linux desktop..."

[Torodung]

"Advance." You keep saying that. I do not think it means what you think it means.

What is needed, before "advancing" anything, is to advance acceptance of the Linux desktop, and IMHO this ain't helping.

Re:Just what we need...

[Anonymous Coward]

I'd rather take one solid piece of software than 10 which are broken in different ways.

Re:Just what we need...

[marcello_dl]

It's interesting how many people are interested in the desktop.

I just care about MY desktop instead. I prefer to edit some configs (which clearly is rocket science) and experiencing some inconsistent look and feel among applications, than using MS or Apple solutions and submit to whatever their management think is a good idea to stick on MY desktop to make their system more difficult to migrate from.

People want to replace xorg (a fork itself)? best wishes, not a problem for me. Problems would arise when xorg replacements are used to introduce incompatibilities, push one distro ahead of the others, render old software/hardware obsolete. It's not easy to pull such stunts by staying free as in GPL, though.

Re:When are they going to use motion sensing and 3

[Anonymous Coward]

It's about time the entire desktop go 3d. It's 2013 and video cards can do it easily. Instead of windows why not just use rotating cubes?

You can easily do that without some fancy display driver or even 3d glasses. Just strap together 6 monitors into a cube shape and fashion a suitable base that will let it rotate in 3 axes (probably best to put the CPU inside the cube so you only need to provide power to the cube). Then to change desktops just flip the cube in the appropriate direction.

While the emerging display servers fight it out...

[Bill_the_Engineer]

While the emerging display servers fight it out, I think I'll just stick to the tried and true X11.

Re:Just what we need...

[lgw]

If an ordinary application can't be written once to look fine on both, then really it is a new flavor. If the application writer doesn't have to know or care, then it's not.

This fork happened over the difficulty in adding candy to the UI, so in theory all the apps that aren't the shell shouldn't care, but in practice, well, time will tell.

Re:Ugh...

[Anonymous Coward]

Meh, bought a MAC

You bought a Media Access Control? How much did it cost?

Explanation

[Alex Belits]

Once Wayland components developers started trying to implement something practical, they discover, one by one, that they need those "unnecessary" X features after all, however there is no way to explain it to the rest of developers, who still believe that removing everything they don't immediately use in their narrow area is a great design practice.

Re:Explanation

[Anonymous Coward]

+1

It's obvious to anyone looking from the outside in.

They should've just went for X11R8, or X12. That is, just fix the real problems. And reset the clock on best practices by discarding the libraries and interfaces that everybody could agree needed to go.

Unix has it's problems, too. Plan9 proved that there was a much better way to do things. Most new features on Linux are half-assed because of the need to maintain some semblance of backwards compatibility. For example, namespaces---Plan9 executed namespaces perfectly; Linux namespaces are an abomination. And yet... they're good enough. Because reinventing the wheel just isn't cost effective.

Open source doesn't help, here, either. Back in the day it made more sense to reinvent everything, because the code for the current state-of-the-art was usually closed source. You usually had to break stuff if you wanted to fix problems.

But in the land of FOSS, you can actually go around and fix everything, and while much less sexy, it's usually the smarter move.

Re:Explanation

[jedidiah]

X is fine locally. It's even more suitable as a gaming platform than Windows and that's probably the best metric for deciding how suitable a local display is.

Remotely, X still manages to hold it's own despite being ancient.

With a few tweaks, X can even kind of keep up with RDP.

The real proof comes when you compare X to MacOS. This is a disaster that you have really experience for yourself to fully appreciate. I think far too many people take all of the pro-Apple propaganda at face value.

Remote MacOS is a disaster and that's the approach the Wayland idiots want to take.

Linux applications don't look like "hammered shit".

That's just mindless Lemming trolling.

Re:Explanation

[Rich0]

The main deficiencies I see with X for remote access are:

1. Applications that insist on client side rendering (maybe some X issue is leading to that, but Chromium is a real pain over a remote connection).
2. It doesn't perform well unless you layer something like NX on top of it. The wire protocol is too chatty or low-level.
3. It needs some kind of middle layer so that you can move applications between displays, and displays between consoles. Think something like screen or tmux. Once you launch an app on a display, it is stuck there.

Wayland obviously isn't going to help with any of this as it currently stands.

Next Woburn/Billerica

[us7892]

I hear that Woburn/Billerica is the next fork of Wayland/Weston, while Wellesley/Southboro is the next fork of Northfield/Norwood. Coming on the heals of Woburn/Billerica is the Provincetown/Gloucester fork. And they're really planning a breakout with a Providence/Cranston fork....

Re:Next Woburn/Billerica

[Intrepid imaginaut]

I'm waiting for the Wayland/Yutani fork myself.

Re:While the emerging display servers fight it out

[ewieling]

This is what people did when Sun's NeWS, Display Postscript, Berlin/Fresco, and Y Window System were released. You are in good company..

Re:Just what we need...

[Anonymous Coward]

I'd rather take one solid piece of software than 10 which are broken in different ways.

I believe this distribution is what you're looking for.

Re:Just what we need...

[lister king of smeg]

I'd rather take one solid piece of software than 10 which are broken in different ways.

I believe this distribution is what you're looking for.

he said solid not unstable or unusable

Northfield + Weston

[AaronLS]

Shouldn't someone create a couple more forks with names like Eastcoast and Southwood so we can have all the cardinal directions covered? Then we can have programmer gang wars.

Re:Just what we need...

[jedidiah]

...yet oddly enough I am still using GNOME2 despite everyone's best attempts to kill it.

A fork needs to be popular to be relevant. Otherwise it is just noise. The same is true of distributions. Few people are aware of them all or even how many there are. Most are highly specialized or have no following to speak of.

So effectively the level of diversity you have to consider yourself is far less than what some Usenet troll might want you to think.

The same is true of display servers. These projects have to gain momentum, developers, and users. Mass revolts may undermine that.

Then there's infighting of course...

I have to confess. I am feeling a big mountain of shadenfreude right now...

You have only worked on trivial problems

[Tenebrousedge]

There comes a point where there is no readily identifiable "best" strategy. Perhaps there are tradeoffs in either direction. Perhaps one persons says, "the rule of thumb that holds for the common case, doesn't apply here." Perhaps there are valid differences about what goal to optimize for -- it is a law of the universe that you can't optimize in all directions at once.

At some point the only way to decide the issue one way is to fork the code and see what becomes popular. As an outsider, you don't really have a good perspective on whether this is justifiable. Clearly the magic code factory has stopped for the moment, but coding efforts are probably stalled more often than not. I started on a new project a few weeks ago, and I don't expect to be doing anything but refactoring and bug fixes for several weeks to come. And if I decided that it was just as much trouble to start over with a bare set of classes and do things the way I think they should have been done the first time, are you going to call me out on it? Is there any better proof of the viability of that strategy but in the execution? Perhaps this will be a better performing or more feature-ful product, and perhaps not, but if the only thing learned from the experience is that "doing it this way turned out to be a bad idea," that still counts as a win in my book.

A failure is something you don't learn anything from.

Lastly, as counterproductive as a fork may be, it's nowhere near as hard to merge changes as it would be if the guy had just started a whole new project. Which is the biggest reason to cry foul over Canonical's development efforts.

Re:Just what we need...

[hawkinspeter]

If he's right, then maybe his fork will gain momentum. If he's wrong, then he'll be wasting some of his time, but I bet he'll learn from the experience either way.

Often it takes someone to branch out and start something to jumpstart development so that other people see what's been done and think it's worth contributing to. Unless someone makes that jump, we'd never know what might have been.

Re:The primary commit history for the past year...

[MBGMorden]

I think the desktop on Linux is truly approaching it's demise.

Actually I think the desktop's only future is on Linux.

Most of the regular users are moving on to tablets and phones. Even in businesses I'm starting to see people migrate to just a tablet. There's still plenty of desktop users but at the rate we're going desktop computers will be a thing of the past within the decade.

I'd wager that geeks and Linux types will be the only ones who still want a desktop OS and system by 2020. We'll probably be running on off the wall hobbyist hardware (Raspberry Pi type devices) and hooking up to mostly HD televisions as monitors if purpose built monitors aren't still available.

I'm not complaining - I'll be keeping mine too as even as a technophile I still prefer to sit down to a full system rather than use a tablet, but I truly think that there will eventually be a "year of Linux on the desktop" - it'll just be after most of the world has forgotten about the desktop.

Re:Just what we need...

[flimflammer]

In fact, find a dr. Sues book

What is that, a book on the benefits of malpractice insurance?