Well, what do we have here? It turns out that Ubuntu isn’t the only Linux distribution who took a left turn off the X.org highway, now driving on a road that will eventually lead to replacing X.org with Wayland. Fedora’s ‘graphics cabal’, as they call themselves, have explained themselves on Fedora’s devel mailing list. They also explain how network transparency can be added to Wayland in a number of different ways, making the mailing list thread intriguing reading material. Also, everybody happy with the headline? No panties in twists this time around…?
Fedora’s ‘graphics cabal’, consisting of Adam Jackson, Kevin Martin, and Dave
Airlie, discussed the whole Wayland thing at last week’s Plumbers conference, and they came to the conclusion that while Wayland is obviously anything but ready for prime time, the eventual advantages of moving to it are so great a switch is pretty much inevitable. Jackson added explicit permission to everyone to quote him, so here we go.
“Wayland’s not a usable default yet,” Jackson stated, “It’ll probably be packaged in F15 as something you can play with. We don’t even have a complete list of transition criteria yet, let alone a timeframe for switching the default. But it’s likely to happen eventually because it’s a serious win for a lot of things, and the downsides are pretty negligible despite the fear from the peanut gallery.”
In a later post, Jackson also handled some of these downsides in more detail, most notably X’ network transparency which is an extremely useful feature for server administrators (and for some desktop users as well, of course). While the protocol doesn’t have network transparency directly, there are options.
“You can still do vnc-like things trivially and with a modest amount of additional wayland protocol (or just inter-client conventions) you can do spice-like things,” he says, “This is good, not bad, because efficient remoting protocols do not look like X. Now we get to design a good one, and in the meantime vnc-style remoting sure does go a long way towards being good enough. (But, we can’t switch yet, because we don’t even have vnc-style remoting yet; so we’re not switching yet.)”
This message from Jackson is mandatory reading, though, as it explains what he means by ‘vnc-like’ and ‘rdp-like’. Since Wayland’s “fundamental object of composition is a whole window”, you could just scrape the pixels out of the buffer and shove them across the network to get per-application remoting. Rdp-like remoting is a bit more involved, but is potentially more useful.
“When I say ‘rdp-like’ I mean ‘install enough awareness of the possibility of remoting in the rendering system that remoting can send a rendering command stream instead of raw pixels if that seems to be a win’. Wordy, I admit,” Jackson details, “And, obviously, much more work than just vnc-like scraping. But it’s a serious win for WAN links, and is the only viable way to remote 3D, etc.”
I’m very happy Fedora is looking to eventually move to Wayland as well, since it’s about time the Linux world looks and moves beyond X to something that has been designed from the ground up to be modern, instead of something that consists of layer upon layer upon layer.
With both Ubuntu and Fedora now on the road to replacing X with Wayland, interest in Wayland will rise. NVIDIA may not have any plans to support it right now, but I’m pretty sure they eventually won’t have much of a choice.
‘rendering command stream’
More and more, it seems every OS eventually becomes Unix, and every windowing system becomes NeWS.
Hardly. Wayland is a lot less like NeWS than X is. This “rendering of command streams” would only apply over the network, and it would be at the toolkit level. Rendering operations are beyond Wayland’s scope.
Oh?
So you’re just going to be the rasteriser to the ‘toolkits’ implementing Pseudo-NeWS.
Very Unix philosophy.
*shrug* I’m more for Emacs than vi (VI VI) personally, and firmly think we’ll eventually be doing nearly all of NeWS in a clumsier fashion in the future.
Plasma’s already halfway there.
http://en.wikipedia.org/wiki/NeWS
I’m 3/4 of the way through ‘The NeWS book’ I think I know what it is.
In case that was just informational, I’ll admit I was a little short.
Just got back from a power outage during which I was reading said book.
Everything NeWS did is still a good idea… It’s amazing that no one’s tried to implement something similar between it and AJAX…
NO, ‘eventually’ should obviously have been entirely in capital letters and UNDERLINED. You are just trying to sneak another one past us you charlatan! I got my eyes on you…
One of Fedora’s major niches has always been playing around with new, possibly not entirely ready for prime time, tech. They did this with the fledgling open source 3d drivers for ati and nvidia for example, projects which are maturing quite nicely in part because of the extra attention they got from Fedora’s early adoption.
As for Ubuntu, I think the community is more annoyed with Shuttleworth and co’s habit of unilateral decisions than the technical issues of moving towards wayland itself.
EDIT: It’s also worth noting that while Wayland isn’t a Red Hat project, it was created by one of Red Hat’s employees
Edited 2010-11-12 23:08 UTC
Yeah, I was thinking of bringing that up in the ubuntu discussion, but I didn’t have time to find the source. Its also being considered for Meego, which is based off of fedora.
Isn’t it Shuttleworth’s right to make whatever decisions he wants? Is his company, and I am pretty sure he is calling the shots. Unilateral is the only real kind of decision in that situation.
You don’t see Apple fans getting annoyed every time Steve makes a decison, and he doesn’t even own his company.
Absolutely. Just as it’s the community’s right to have opinions on those decisions. Microsoft and Apple both get more flak on a regular basis than this tempest in a teacup Ubuntu is experiencing
My problem wasn’t so much with Wayland, but with Canonical and Mark’s statements. He basically claimed that it was impossible to create a snazzy desktop app with X which is absurd. X will do everything that Windows or OS X can do and then some. I also have a problem with Canonical, which commits very little code to the community (this doesn’t belittle other contributions) making grandiose statements about the future of the architecture. Its not like they are going to do the work themselves (at least the coding and testing). But I admit I am biased.
X is a very powerful toolkit… but large and somewhat complicated (I think)… but it has served well for several decades.
Well, actually I don’t think that’ what he said at all. He said that it was hard to make a snazzy desktop with X without hacking it to pieces or at least alluded to that. It can be done but at what point does the effort become counter intuitive? Just look at all the hacks that had to be done just to get a decent graphical boot. It took YEARS for X to get to this point, and while appreciate the work that the devs have done (which make Wayland possible, btw) the situation require a lot of hacks on the distro makers part just to get things to look decent.
Wayland uses KMS, what that means is that potentially can wayland load after your kernel does and drop you into your login manager (GDM, KDM) while your machine loads the rest of you services in the background. Ubuntu does this now in-fact except it has to run the graphical boot, loads X in the background and switches to it when lt’s loaded and continues loading processes in the background. With Wayland they can skip loading X altogether and get even faster boot times and a better user experience. The display server is already loaded you can skip a lot of things that distros have to do now because they have to load X. If you’ve ever booted a PS3 or PSP you will see what I’m getting at. You can potentially drop the user into their desktop within seconds all seamless, no context switches, no graphical boot loader, no X loading.
The second paragraph in your post shows that you don’t understand most of the concepts that you were referencing.
“X will do everything that Windows or OS X can do and then some.”
I think you’re missing the point. The point of X is to allow X clients to run. X clients can communicate with any “server” that matches their network protocol.
As such, instead of making the primary display server an “X” server first and foremost, why doesn’t it be a graphics and compositing server with an “X” server protocol layer that can run X clients?
This would make the most sense, and is exactly the way Mac OS X does it – and it’s how Wayland wants to do it.
X is not a desktop technology – although it can be used as such (clunky, and hacky), it’s a technology to display graphical apps over a network (clumsy in today’s PCI-Express X16-2 high-bandwidth world)
Jump on the bandwagon man – it makes the most sense by far and it’s long overdue if *NIX ever wants to compete from a raw graphics capability perspective.
X is not what you think it is.
It is a windowing system first, and the network transparency comes second. Not the other way around.
X can do exactly what any “modern” windowing system found in Windows and Mac, and more. Which is why it will be probably be forever damned to be misunderstood.
What X needs is an actual standardized toolkit, and a consistent way of doing things. The problem of X is that it was open enough that there ended being offered more than 2 ways of doing the same, almost it seems through its entire history: Athena vs. Motif vs. OpenLook then, Gnome vs. KDE vs. whatever now, GTK vs. Qt, etc. etc. etc.
The problem is not X really, but what is built on top of it. All it is necessary is simply to build, and ENFORCE, a common set of standards for the widget set, window manager, and consistent look and feel of apps. The route Ubuntu et al are proposing involves all that PLUS reinventing the wheel for the windowing/display stack. I do not think that is necessarily the best solution. It will take longer and more effort… but whatever, some people do not enjoy efficiency and crave making things harder that they need to be.
Edited 2010-11-13 09:52 UTC
I have to disagree.
In my mind, X is foremost a method of drawing 2d primitives in a hardware- and location-agnostic manner. WMs such as Enlightenment, FVWM, XFCE, Gnome, and KDE provide a windowing system. QT and GTK provide an abstracted form of drawing common UI elements via X primitives.
I love X, mainly because I access Linux and BSD systems from my Windows desktop, and use a Windows X server to display graphical apps along side my other apps on my Windows desktop. When running on a desktop system that is running X natively, it’s flexibility is wonderful, as it allows me to run my desktop how I like, instead of in a manner built around somebody else’s idea on how a UI should be.
The fact that X does is not a windowing system allows X apps to operate within a Windows environment without much trouble. Sure, load/save dialogs look weird, but the main windowing methods provided by Windows work the same, whether they are X apps or Windows apps.
The problem with X is that it suffered for a long time with lackluster development. It was not quick to adapt to new hardware paradigms, and at the same time graphics hardware makers didn’t keep it a first-class citizen.
When X was a first-class citizen for Unix system makers, it was a top-notch solution. The first complaints about X performance I ever heard were from Sun customers, and that was after Sun begain using ATI’s Rage Pro hardware for low-end graphics in their lower cost desktops. This wasn’t due to inherent limitations in X, but rather PC graphics hardware not accelerating X-style drawing primitives efficiently, as they weren’t necessary for Windows performance.
As it is, using XMing to display *nix software (running in a virtual machine) is usually higher performing than other methods, with the added benefit of apps fitting well alongside native local applications.
Wayland, as it is, appears as a “me too” solution, and that is what is not needed for *nix systems.
Wayland isn’t a “me too” solution. It’s goals are completely different than “X” proper.
The goal of Wayland is a complete graphics hardware stack with a set of core capabilities like OpenGL ES (required), compositing (required), etc.
X is a technology to display clients – that may or may not be running on your box. The various servers aren’t the proper solution for advanced graphics capabilities without a billion extensions that may or may not be present.
Wayland is they way all major solutions are doing it nowadays.
It may be time to be open to the fact that 1980’s-designed technology might be showing it’s age in modern times.
That’s what I meant as a “me too” solution. I wasn’t comparing it to X, but rather Windows and MacOS X. It is something that adopts similar goals and similar solutions to attain those goals, rather than a different set of design goals that X embodies.
I think technically, you were sort of agreeing…
If you look at how many holes xorg has pushed into the basic X implementation to get modern stuff running then you can see that there is a serious problem, also given the speed of development, I guess the entire concept is way overdesigned. X was the lesser option compared to Display PDF and NeWS back then and still scale not as well as other systems, not even on the network, where RDP runs circles around remoted X and even the mac2mac remoting runs performs better than X11.
“X can do exactly what any “modern” windowing system found in Windows and Mac, and more.” –> Please define X in this case. Xorg server? XFree86? X is a protocol (API) for displaying clients, not a desktop technology. If X was a desktop technology, one would figure it would have a hard time sharing the same space as Mac OS X desktop and Windows Desktop through Gygwin.
In addition, “X” – I’ll define it for you… Xorg server… can do most things that Windows and Mac OS X can do through a handful of extensions that may or may not be loaded which is impossible to bet on when programming a desktop environment – like KDE and Gnome. I applaud the FOSS community for taking Xorg as far as they can.
The last time I looked, DRI wasn’t really intended to be a part of “X” proper. *extension… might be there… if you’re lucky.
Gee wiz, given the context of this article… obviously I must have been talking about X running DecWindows on a VAX.
As of today, X refers mainly to X.org for the desktop windowing system side of things. Given that most non x86 unix workstations have gone the way of the dodo.
GDE and KDE are exactly what it is wrong with the FOSS approach: two ways of doing the same thing, which have taken twice as long reinventing the wheel to achieve basically the same result.
People who focus on the warts of x.org are completely missing that big picture. Reinventing the windowing system/technology if it is going to lead to another multitude of different ways of doing the exact same thing, is going to put them exactly where they began. Again.
I wasn’t referring to what platform by requesting “Please define X?”, I was referring to what X as in what layer… X proper the spec? X with DRI extension? Xorg the server? XFree86 the server, etc.?
I agree in the sense that GDE and KDE are competing and reinventing the wheel. There is so much potential there if they weren’t writing 2 calculators, 2 notepads, etc.
I think Wayland is different in the sense that the Vista/7 DWM is different as opposed to XP’s windowing system while still supporting XP-coded clients. It’s a successor, not a competitor.
I think that if Wayland succeeds, Xorg will be considered old hat on GNU/Linux systems. Maybe not additional *NIX that have different kernel capabilities, but certainly on Linux.
What say you?
Without a canonical ‘fallback’ standard toolkit, I kind of agree…
I think Qt4 is a better library (more capable, flexible, fast) but uses a touch more memory that I’d like.
GTK(2(+)) just makes me sigh… Like GNOME, it seems like a bunch of disparate elements munged together.
Frankly, I think computing needs to go tabula rasa on the whole OS deal…
Everything’s Unix-like on some philosophical level now. (W7 is the least Unix-like of the big OSen, but compare it with the structure XP/98!) but we still have to port/maintain everything over and over…
I’m not sure what the solution is, but I have a very vague idea….
Then please share that idea with us
I don’t know about others, but myself I’m not sure I understand what you’re talking about.
Windows, OSX, and Unices are too deeply entrenched in their legacy to make breakthrough changes nowadays. They can clean up some dusty components, but anytime they do they must reimplement all existing functionality on top of it for compatibility reason, so the process is tedious, lengthy and error-prone. See KDE 4.
In my opinion, the most customers a piece of software has, the smaller its evolution capabilities are due to the need to please them. Most extreme example being Haiku : the project does not even have a stable release, yet anyone wanting to break compatibility with BeOS for good reasons would be met with extreme hostility.
The only way to do some deep work on the desktop would be to write a new OS from the ground up, I think, and it’ll take ages before the thing catches up, except if it manages to fill a niche first…
Edited 2010-11-14 08:39 UTC
I have ideas, but I know that my ideas aren’t complete enough to spread around.
No need to spout bullshit while I’m working things out, and yes, I did mean fully writing a new OS.
Every new RTOS being a mini Unix is…boring.
Change isn’t easy, but I think it will need to happen.
Well, I’m working on an OS project as a hobby, and I think I understand. In the current state of the project, I start to have a nice stack of theoretical work done. But before I get more people excited about this than me and the two guys who post a comment on my blog from times to times, I need much, much more working code than in the current codebase.
Good luck on your side !
Edited 2010-11-14 09:56 UTC
Heh.
This will be years before we can call what we’re doing an OS, rather than a collection of libraries.
It’s really ambitious, as it’s basically a language-based system, that has 4 languages with a common message-passing system, and resolution-independence and _display type_ independence as key concerns. (a program that runs on a bitmap display should work on a text display, with (significantly) less detail, and likely reduced functionality)
I’m getting close to having enough worked out to write a spec down to the api, but it’s all theoretical still. I want to make sure that things all make sense before we go past finding/vetting libraries.
There’s so much code out there that’s published as ‘license-free software’, which takes permission to re-license as free… Luckily it’s often old enough that the author’s virtually forgotten about it, and just gives it to you.
Well, funny, I’ve been thinking about resolution independence and display type independence too.
My advice, if you did not thought about that already : when studying resolution independence, consider not only display resolution but also input resolution. Applications automatically adapting themselves to mouse, stylus, or touchscreen input by resizing controls would be something awesome, in my opinion.
Edited 2010-11-14 18:03 UTC
Oh, absolutely.
Considering we’ll be basically re-implementing NeWS (with a number of changes for multiprocessor, 3D acceleration and other modern features) to the point of using ps, using processes similar to font kerning on interfaces and interface elements makes sense to me.
The rate at which buttons shrink relative to each other, what elements remain once you drop to a certain resolution…
It really comes down to setting the priority of elements and visual attributes of them, then determining how they change at different resolutions/bit depths/available line elements.
You realize that the guy who started Wayland – Kristian H~A,gsberg – has been working on X[1] on a daily basis for some time ?
Reading the initial mail, the guys pushing for Wayland in Fedora include David Airlie[2], who is working on the open source radeon drivers. Adam Jackons (ajax)[3] is also one of them and he has also been working on various X stuff for a long time.
Not to say that you should necessarily share their opinions because they are working on X. But they probably have *very good* reasons to push for a replacement to the software they’ve been working on for years.
And if they say that the X protocol has drawbacks (multiple communications for a single mouse event, unavoidable flickering and so on), they probably don’t just made it up… they’ve been working on it for years.
[1] His nickname is “krh” : http://cgit.freedesktop.org/xorg/
[2] http://airlied.livejournal.com/
[3] http://people.freedesktop.org/~ajax/
I understand where they are coming from, I simply disagree that starting from scratch and reinventing the wheel over again is the best approach.
BTW, this is not the first time that “someone working on X daily” has decided that X is brain dead and knows exactly how to fix things with an alternative, which eventually… goes… nowhere.
I have seen this movie before.
Edited 2010-11-13 23:13 UTC
[citation needed]
/me wonders which full-time X developers did this.
In this case you haven’t seen shit, you are talking about not just one X developer, you are talking about the main contributors to X development. The people who realised we needed to change years ago, instigated projects like KMS and GEM/TTM drivers, in order to create an alternative to the X stack in the future.
We aren’t mouthpieces like spaceboy, we are the people in the trenches, we just don’t come out spewing unrealistic timelines, wayland is in the future, but when the future becomes current is anyones guess.
I have no problem with X developers talking about what’s wrong with X and what we’ll do about it. X will not and should not last forever; the time for change has certainly come. The new system needs to be well planned and considered such that it can last another 20 years, but presuming that this is done I have no problems with it.
What I don’t like are people who are not X hackers and who know nothing about its history and function declaring that X is dead and hailing the great savior Wayland, as if all things they perceive as wrong with Linux desktop environments will magically disappear when Wayland descends from the heavens.
I can’t mod you up enough.
This has happened before…many times. I remember when DirectFB was going to be the next big thing and it would solve all of our problems. Then it was “Y” and other small projects that faded into oblivion. Now it’s Wayland. Wayland certainly has a better chance, given its progenitors but it is still a long, long way off before it even has the potential to overtake X. By then X may have evolved into a very different beast, and/or Wayland may become insufficient for future needs.
I think that the last statement is likely. I’m really not convinced that GL is good enough for a windowing system. one more thing that assumes I’m writing C.
On a totally related note we are now left with exactly two viable choices for a name for an X successor: X12 and Z. The fact that wayland starts with W clearly indicates that is an evolutionary step backwards, since X was better than W from day one (alphabetically speaking).
Apparently Wayland was developed for meego, where they are going to drop X11 in favor of wayland probably due to the leaner memory footprint and better performance.
“It’s also worth noting that while Wayland isn’t a Red Hat project, it was created by one of Red Hat’s employees”
If I’m not mistaken, that developer now works for Intel.
panties straight over here, nice job on that one, Thom.
We can’t have a thread on the X-server without somebody complaining about misleading headlines.
Claiming that Fedora will eventually switch to Wayland is a tad more than the statements can handle. It would be more correct to say “Fedora-developers toying with the idea of an optional inclusion of Wayland in Fedora 15 and probably sometime in the very distant future making Wayland the default – while keeping X.org as fallback”.
However, that’s not a good headline, is it?
And Thom? Sowwyyyyhhh! :p
Even better headline: “Four pissed developpers claim Wayland is the way to go, flamewars ensue”
I would hardly call it a flamewar. At least not the mails I’ve read. There’s a few worried voices, but that’s it. We know Wayland has won when it goes stable in portage. That’ll take some years…
You just might be true. The sad thing is that volume might won over function in this issue (i.e. if distros with large user basestake it up). Network transparency _is_ a major functionality of X, and as also some people in the Fedore discussion thread have said, a new framework can’t replicate this functionality well if it’s not included in the design considerations from the first steps. Adding rdp/vnc functionality on top is a far worse solution, coming from trials to replicate X’s networking capabilities in systems that lacked them. If they just want to create a new branch of Linux based desktop systems based on only client/desktop functionality, that should be acceptable, but that should be a _clear_ and loud message from the start, and a generial replacement of X should not be a goal. And maybe it isn’t, but then their communication sucks.
Anyway looking at the Wayland architecture text, Wayland doesn’t seem to want to replace X server, it just wants to sit below the X server, eventually making X performance worse (this is just a feeling, I don’t know that for sure), But still, from all the text and graphics, it seems that the “new X” on top of Wayland would be a reduced version of the current X and network capabilities are never mentioned anywhere.
I think we just need someone to clear the air about Wayland’s exact goals and its stance towards X. Speculation has run its course enough, I’d say.
The problem with X11s performance on top of wayland is not given because if you run it locally the toolkits will bypass X11 altogether and for remoteing face it the rendering performance for remoting is the smallest bottleneck the biggest one is the dreadful x11 protocol itself.
Besides that:
I am not sure if RDP is really a worse solution than X11, after all it performs way better than X11 ever did.
The thread is pretty clear, on one hand are the guys who want to do it, basically also X11 contributors, on the other hand a bunch of admins who fear to loose their X11 remoting, which never really worked well (and the X11 guys pretty clearly stated that even a single button press causes multiple roundtrips) and they offered various solutions, but they also said, network transparency should not be part of the rendering core because otherwise you will end up in the same mess X11 nowadays is regarding modern uis, it depends on whether it will be solved like a vnc on windowing level or a drawing level. But there will be a transitional period where both will work until this issue is resolved. But in the end we will get something better working than the slow as molasses network stuffing X11 protocol!
True enough – the only thing that thread says is that someone believes Fedora will probably move to Wayland at some point in the future. Breaking News!!!
To have a more development-oriented distro like Fedora (rather than Ubuntu, which is generally accused of not doing much in the way of code) going to Wayland in addition to Ubuntu could give it quite a bit of momentum. Obviously, there’s still a lot to be ironed out, and we’ll have to see how this goes, but this definitely seems positive.
The graphics stack on Linux needs a major jolt in the arm, and if Wayland is the way to do that, then I’m all for it. I dream of the day that graphics on Linux just work and work well. In many ways, it’s better than it used to be, but we need something way better than we have, and having two major distros behind Wayland should give it quite a bit of traction.
Of course, as fantastic as this news is, it’ll be a while before it really bears fruit, but you to start somewhere.
This is hardly surprising since Wayland is a Red Hat engineers project… I have considered it somewhat inevitable since GEM and KMS went into the kernel that something like Wayland would come about. I’m not even entirely convinced this qualifies as news, heh.
Wayland is something closer to “proof of concept” rather than any alternative. Also X isn’t going to stand still while Wayland is readied for prime time. I’m all for improvements, but like most things, the proposal is easier than the execution.
Also, layers are not only not bad, they are something which makes a great deal of things more modular and easier to adapt and debug. Do you hate TCP/IP? That is at least 4 layers Phy/IP/TCP/App. Each layer does one thing well.
X has solved problems which Wayland now has to. How do you do all the devices? Make a (wacom) tablet or touch screen act like a mouse? Support N buttons where N can have a large range? Work on embedded graphics as well as 3D hardware – when they are different chips like on newer laptops? The drivers and API will have to go somewhere. And the apps will have to run on top of it – does it have every capability now used by Gnome and KDE?
You can do things with the framebuffer now, but not the desktop.
I hope Wayland succeeds and is better, but I fear that
1. It might take a few more years than everyone is thinking to get something like LibreOffice to full function.
2. It won’t support a lot of hardware (legacy or specialized) – and consider accessibility or overlays.
3. It might be limited or require more resources by the time it is finished.
4. What about nonfree apps like Google Earth? Will they switch or have two versions or will there be a compatibility lib (on either side and one that doesn’t slow everything)?
Actually, I have seen pictures of Wayland running with 2 X servers running in it. If it runs below X so that X and X apps don’t have to change, that might be a cool thing. It does require more graphics hardware than X does though, so I doubt it will ever be the default on a server distro. Of course, in 5 years maybe all servers will have 3d graphics on board.
I’ve never understood why people would want a GUI on a server.
actually, it is exactly because they don’t want it. strange but true.
you set server into runlevel 3. and when you need some software you simply do this from remote machine:
ssh username@machine -Y whatever_command (like. firefox, system-config-network)
and graphic session runs on local machine not server. server is still in runlevel 3. and you never needed to go into runlevel 5 on server to configure something from gui.
i’m avid network X addict, but as long as wayland options me to run seamless remote session like one before while server is in runlevel 3… i’m all for it. seamless remoting shouldn’t be hard to implement with any protocol,… while in runlevel 3? well, that i don’t know
Edited 2010-11-13 17:10 UTC
What is it about runlevel 3 that makes you think it can’t run X? Because Red Hat said so?
On my system runlevel 2 is normal and runlevel 3 starts up some optional stuff I don’t need all the time. Runlevels 4 and 5 I don’t even bother with.
Beware the fallacy of presuming that how you do things is the universal standard. That’s bad in a Windowsy sort of way.
You don’t need GUI – far from it, all you do need X libraries (inc/ tool-kits such as GTK/QT) and SSHD to run your favorite administration tool.
Better yet, add SSH compression (ssh -C) to the mix and you can actually run X application over a fairly slow DSL or cable line. (Though I would agree that remote X is not a speed daemon – especially when uplink is below 512KB)
– Gilboa
Because GPU computing has potential?
X isn’t needed for that, at least there’s no reason it should be.
Wayland aims to be an alternative, however, you appear to be saying don’t bother with Wayland cuz its hard? Thing you appear to be stuck on is that most Linux installations today are on form factors where they simply do not care about remote displays etc, and the only other use is servers which do not – and imo should not – need a GUI. If you need idiot-proof interfaces to manage your server, things like WebYast are around already.
I simply think you vastly over-estimate the usefulness of enabling network transparency in the display server. Also, Ubuntu has already committed to getting Utouch working with Wayland, so that concern needn’t be maintained either.
I wouldn’t say so. Having X on the server, providing remote X access to clients doesn’t mean the server “needs” or uses a GUI. It means providing quality remote access to remote X clients even on limited bandwidth connections, and I wouldn’t want to see that go, or be replaced by junk rdp/vnc capabilities.
>”Also, everybody happy with the headline? No panties in twists this time around…?”
The headline is OK, but not the unprofessional and childish remarks quoted above. Imagine reading a newspaper and you came across something like that.
I don’t think I’ll both coming here any more if even the editors here write such childish articles.
And this is exactly the whiningism that he is talking about with his “panties in a twist” comment. You could’ve just said, “why yes, I do have my panties in a twist again” and been done with it.
Your choice–but I actually like Thom’s style of writing. If I wanted to read something dull and boring, er, I mean “professional,” I would read a newspaper. Thankfully there’s something less bland than that out there.
“The headline is OK, but not the unprofessional and childish remarks quoted above.”
Translation –> “Hi, I’m a grump.”
“Panties in twists” is something called humor. Humor is sometimes a great way of making a point without direct confrontation. It’s used by professionals all the time to disarm a potentially hostile audience.
Maybe you need visited by Bob the Dinosaur?
Edited 2010-11-13 17:12 UTC
I’ve said it once and I’ll say it again:
Don’t let the door hit you in the ass on the way out…
Well, then don’t. This is the internet and not a newspaper. I found the remark rather funny. I prefer some personality in my blogs.
Also, everybody happy with the headline?
Aside from the fact that you split an infinitive?
Ha ha!
[edit]
On the other hand, English is probably his 3rd language.
The only other languages I know are Spanish and Latin, and I’ll be damned if I can speak or write them fluidly. I did once get drunk with and had a 2 hour discussion with a french-speaking man.
Edited 2010-11-13 04:20 UTC
English guides no longer discourage the use of the split infinitive. The objection was tied to history and not logic.
Why? Because on top of all the braindead “WTF?” decisions Ubuntu is making with every single version of their distro, they’re even making Unity default on desktops. Seriously, WTF?! As if the removal of the GIMP by default with no decent replacement, disabling of Ctrl+Alt+Backspace (with no simple way to reverse this), moving the window manager buttons to the left (again with no way to reverse aside from gconf-editor), and countless other retarded and/or premature changes weren’t enough… let’s just add Wayland to that list of ridiculous changes.
At least Fedora so far is just replacing X with a modern alternative, not replacing the whole damn desktop environment with something that was designed for a damn netbook on top of that. Ironically, I completely gave up on Fedora a while back partially because of their shipping of software that was just too bleeding edge and buggy, and in some cases even still under heavy development and not yet ready for general use (ie. KDE4). Up until the last several versions Ubuntu was doing alright–but now, they’re pulling the same crap Fedora is known for in a way.
I won’t be one of the first to hop onboard, but I am anxious to try it out in a virtual machine and see how it turns out. This should be interesting.
Do not think it is a small project to replace X with Wayland, it is going to take years.
No, no panties this time..
I feel a little bit better, now that Red Hat recognizes the importance of remote displays, but network transparency makes X special, instead of a “me too” solution that Wayland seems to offer.
Its fine to have the choice of either. Some people think that the majority of people don’t need network transparency, and maybe they are right. But X is one of the things that goes into making Linux a truly multi user system as opposed to Windows or OS X. Thats important for a lot of people. My only fear is that support of X will suffer because everyone want to work on the new shiny.
Between now, and when Wayland is looked at seems like at least 1 year.
A lot can happen in a year.
Xorg may change to meet the challenges or requirements of Wayland.
I think most people who think Wayland may be the way to go are discounting the fact Xorg may implement alternatives that make Wayland much less attractive.
Besides, Xorg has a huge installed base.
The benefits of Wayland will have to be extraordinary for Xorg to be dumped.
While those benefits are analysed by the X community, we may find some sort of hybrid come out of Waylands ideas and Xorg’s current standards.
It _is_ exciting to see a competing standard though developing in the Linux space.
It will be interesting to see if Xorg has the engineering stamina to sustain, repel absorb for fall against Wayland ideas.
-Hack
As excited as I am about the project I have to ask where this will put non-Linux operating systems which rely on Xorg as a display server? The open source world is more than just Linux, it includes a whole ecosystem of different operating systems bound together through the ability to share projects and code.
If a significant number of programmers peel off from Xorg then what will happen to the open source world? we’ll end up seeing the development of a monoculture that is even worse than it exists today – right now there are just the annoyances of gcc’isms and GNU’isms but the move away from Xorg to something bound to Linux demonstrates something of a destructive move against plurality.
I really do honestly like the idea of Wayland but the programmers have failed to demonstrate that this is actually more than just a ‘Linux thing’ given how reliant it is on features within the Linux kernel to operate. For me I prefer *BSD but if my choice is limited because a good amount of programmers are neglecting Xorg is it really beneficial to the wider open source movement for such a concentration around a single operating system core?
This ecosystem is quite often a one-way stream, the contributions to X.org from non-Linux (any one OpenSolaris developer) is very very low, *BSD users expect because someone did some work 5-10 years ago to make X.org work that it should continue to work.
The XFree86 driver model has always been a hack, fixing it by putting the proper bits into the OS kernel was how things worked back when X was originally designed, XFree86 was a diversion in the road that held real development back in the name of cross-OS insanity.
So really all the technologies are open, all the code is suitable cross-licensed, all it takes is the people who care about the other open OSes to actually care enough to make a real time commitment to porting the code. X.org already relies on the Linux kernel to get features people want, until other OS kernels can keep up they will remain stuck in 1995.
Not sure what you mean by GNU’isms (please explain), but from your earlier posts I think that gcc’isms are GCC extensions, and that they annoy you because many of them fail to compile with say Clang.
The thing is that these extensions where added mainly at the behest of Linus and the other Linux kernel developers.
Linus has stated his displeasure with GCC devs being reluctant to add extensions since they want to adhere to the standards. So you are pretty much barking up the wrong tree here.
On the other hand, the Linux kernel devs probably don’t ask for extensions on a whim, they know what they want and how it will improve their productivity/solve problems. That other devs outside of the kernel space find these extensions beneficiary and use them aswell supports this notion (and this would likely be the root of your problem, since I doubt you are trying to compile the linux kernel).
So next time you try to compile something from ports using Clang and it fails, know that it’s not due to some evil scheme of the GCC devs. These extensions have been added through pressure from programmers, who in turn obviously will make use of them.
As long as there is decent network transparency, I’m in. I’ve not heard of a xnest/Xephyr equivalent, but I’m sure there can be. As long as we aren’t throwing the good out with the bad, it’s a win and a worthwhile clean up.
All the problems of X and Wayland have already been solved 30 years ago with the NeWS Window System:
http://en.wikipedia.org/wiki/NeWS
I am not saying that NeWS should be brought back from the dead, but it can be used as a design guide because it solves the problems of the X Window System very elegantly.