“X.Org Server 1.11 was officially released this Friday evening. X.Org Server 1.11 was originally planned for released in mid August, but then the unfortunate passing of Keith Packard’s mother (the X.Org release manager) led to a one-week delay. Our condolences go out to Keith Packard and his family. After numerous belated releases from X.Org in the past, a one week delay is nothing to complain about, especially considering the sad circumstance. While this is a new major X.Org Server release, it’s mostly about bug-fixing. X Input 2.1 was delayed (with its touch-related features) to the next X.Org Server release (or later), there isn’t any RandR extensions (after RandR 1.4 was restarted), and just nothing to get too excited over, besides addressing outstanding issues. Regardless, it’s an improvement that incorporates six months of enhancements.”
Thom, I’m not complaining, but I’m just curious why this release merits a mention in particular. Any reason?
It was submitted?
I don’t follow X development, but if someone submits it, why not mention it once in a while?
Fair enough!
Maybe its the last interesting release, before every distro is switching to wayland 😀
I hope so, I really want to see Wayland being adopted already.
Phoronix (who do follow X development) had an article about the current state of Wayland in Ubuntu not long ago. Short version: it’s useless in its current state.
I’m sure it will come along sooner or later, though.
I’ve been following Wayland very closely and I don’t think it’s useless. Sure it’s not usable right now but I also think it’s far from being useless. I mean, most open source drivers already work with it (nouveau, intel, radeon).
Xorg can run as a Wayland client. GTK+ 3 and Qt 4.8/5.0 also has native Wayland support. KWin is currently being refactored to have Wayland support.
We only need window managers to be ported to Wayland now, so we can start using it.
Edited 2011-08-28 23:43 UTC
We really need nvidia to make a Wayland driver.
AFAIK, Wayland depends on kernel modesetting (KMS) and on graphics memory management in the kernel.
Kernel developers will not allow binary code to be included in the kernel, as it violates the license (the exception allowed is for binary blobs that are firmware to be loaded into hardware, such as for wireless adaptors, which is not part of the kernel itself).
nVidia’s binary blob graphics driver gets around this by nVidia writing an open source kernel loadable module wrapper, which interfaces to the kernel, and also to the binary blob. A two-way bridge, if you like.
Anyway, for now, nVidia cannot make a proprietary closed-source Wayland driver, because it won’t be accepted by the Linux kernel developers to to have KMS and a graphics memory manager as open source and part of the kernel when the main piece of the driver is a loadable module and a binary blob. AFAIK this just won’t work.
Nvidia have said they have no plans to make a Wayland driver.
Edited 2011-08-29 07:09 UTC
I read a developer on #wayland have said that Wayland doesn’t really require KMS.
AFAIK it did have that dependency to start with, I don’t know what the status is now.
Nevertheless, AFAIK, nVidia have no plans to support Wayland, and the Linux kernel developers have no plans to accomodate unfixable binary blobs from nvidia as part of the kernel.
http://www.phoronix.com/scan.php?page=news_item&px=ODc2Mg
NVIDIA Says It Has No Plans To Support Wayland
PS: Note that the open source drivers from Xorg for AMD/ATI GPUs are part of the Linux kernel, do support KMS and GEM, and hence will work with Wayland.
Edited 2011-08-29 07:24 UTC
My opinion is that they will eventually have to support it. If they want to remain relevant on Linux at least. As the whole Linux ecosystem is moving to Wayland as we speak.
Then we simply have different conceptions of usability. Not being usable is exactly what I mean by ‘useless’.
Doubt it. Got a feeling Wayland is going to be as much of a mess as Gnome 3.
Edited 2011-08-29 00:10 UTC
Doubt it. Got a feeling Wayland is going to be as much of a mess as Gnome 3. [/q]
A mess for people like you who are afraid of change maybe.
Those people who want a modern and smooth desktop welcome and see Wayland as a great thing, me included.
Edited 2011-08-29 00:40 UTC
Duplicated reply, removed.
Edited 2011-08-29 08:12 UTC
Hehe. Again, people are NOT afraid of changes in general. People are only afraid of BAD changes, e.g. from working to unemployment, from live to death, from being healthy to having heartattack.
I agree with you on that. Fear is a natural thing, and I have my own fears too (which are not related to Wayland at all).
But if we look at Wayland, it seems to be a good thing, a more simple architecture, clean code base, direct access to the compositor, ability for clients to control the rendering themselves enough that we never see tearing, flicker, lag, or redrawing problems.
Why fear this or call it a “mess”? it’s not like we are going backwards. To me, staying forever with X is going backwards and a mess.
Edited 2011-08-29 08:39 UTC
X—and I cannot emphasize this enough—X works. Does it work optimally for all cases? Of course not. But for nearly all uses, it’s fine. The only “going backwards” would be switching to Wayland before it works in at least as many places as X. Or, to put it another way, that would be PulseAudio come again.
It wont be PulseAudio all over again. And I’m sure the transition will be well made, most of the Wayland developers are also X developers, so no need to worry about it.
Wayland is a composting display server. Not sure what does Gnome’s mess have to do with it?
Dumb question. What happened to XFree86?
They mucked with the license. Nobody likes that.
Little bit more information, from Wikipedia:
“In February 2004, with version 4.4.0, The XFree86 Project adopted a license change that the Free Software Foundation considered GPL incompatible. Most Linux distributions found the potential legal issues unacceptable and moved to a fork from before the license change. The first fork was the abortive Xouvert, but X.Org Server soon became dominant. Most XFree86 developers, who were already annoyed at other issues in the project, also moved to X.Org. The last CVS commit was February 2009.
[snip]
2004: Licensing controversy
Versions of XFree86 up to and including some release candidates for 4.4.0 were under the MIT License, a permissive, non-copyleft free software license. XFree86 4.4 was released in February 2004 with a change to the license: the addition of a credit clause, similar to that in the original BSD license, but broader in scope. Many projects relying on XFree86 found the new license unacceptable, and the Free Software Foundation considers it incompatible with the version 2 of the GNU General Public License, though compatible with version 3. The XFree86 Project states that the license is “as GPL compatible as any and all previous versions were”, but does not mention which version or versions of the GPL this is valid for.
Some projects made releases (notably OpenBSD 3.5 and 3.6, and Debian 3.1 “Sarge”) based on XFree86 version 4.4 RC2, the last version under the old license. Most operating systems incorporating XFree86 (including later versions of OpenBSD and Debian) migrated to the X.Org Server.”
This will probably break binary drivers for no reason.
ABI compatibility should not be set in stone.
If that means vendors have to recompile their drivers, so be it.
Breaking compatibility for what is essentially a bugfix is just stupid IMO.
Accepting obscured code which can bring down your whole product, without having any insight into what is in it or how it works, and consequently having no ability to fix it if your users complain, is just stupid IMO.
You really live in a naive world, don’t you?
People, who explicitly want Linux desktop (me included) can make an effort to get OSS supported hardware and with inevitable standardization of basic video HW (with the arrival of GPU enabled CPUs like SNB, IVB and LLiano) that should be trivial.
For others Linux PC desktop has long been irrelevant anyway. By the time Wayland becomes ubiquitous, not much people would miss nvidia PC linux support. At the same time nvidia will fully support their mobile chips on linux where it actually makes any impact.
What a way to go off topic, “Oh, but Wayland is coming and it’s going to make everything better!” I’m not impressed, what impresses me is how versatile X is. Going with Wayland will remove some of that versatility.
Network transparency is truly useful. Sure not to the majority of users out there, but then the majority of users feel that X.Org’s performance is good enough.
I’m under that category, besides, if you try out a Gnome 2 distro like Debian Squeeze, or Scientific Linux on some Core 2 processor (not even the current generation, or even less than that) you can feel how snappy it is compared to Windows 7. This is even without any compositing window manager.
Why would we drop all the driver compatibility just to get some more speed? I also understand Wayland is supposed to be cleaner, but X.Org has helped a lot by moving to a more modular way of doing things.
Either way, I see Wayland as a “well we want improvements, but we don’t want to help the project that is out there ’cause we want to write our own!” mentality.
This mentality can be good, but really isn’t when you’re talking about something that is to most things as important as the kernel itself. Actually at this point in time, the kernel is even less important than X, since we have alternatives to the Linux kernel.
Wayland is not really duplicating or dropping any work. Check this from their FAQ:
http://wayland.freedesktop.org/faq.html
Wayland basically takes everything that works from Xorg and throws away what doesn’t, it does also use existing kernel components and it simplifies the architecture for better performance and removing latency. Network transparency hasn’t been defined yet, but I’m sure it will be there. I heard somebody was working on that on #wayland.
Edited 2011-08-29 19:45 UTC
There was a GSOC to add network transparency to Wayland: it failed.
This doesn’t mean that network transparency is impossible to add to Wayland, just that it isn’t simple.
BTW, I disagree with you (and quoting the Wayland FAQ is quite useless): Wayland re-invent a lot of things which wouldn’t have been necessary if the ‘GPU-memory sharing’ had been added as an option in X12 (X12=X11 removing all the obsolete options+GPU memory sharing option) instead of a separate project.
Would people please shut up about network transparency. Yes, it is useful…
but network transparency as handled by X11 is really antiquated. None of the toolkits draw with X11 any more so the way things should be handled these days is that the toolkits should handle remote drawing.
Either way you can run x within wayland for all the applications that are not ported (much like Xquartz).
I suggest you spin up that cpu with a heavy load and then check out the amazing snappiness. the thing here is that without compositing you’re using the cpu for graphics, which is far more inefficient than using the gpu. Compositing isn’t supposed to make things “snappier” just improve the experience (like not spinning up the cpu when drag gin windows etc).
that isn’t really an accurate characterization of the project. It is more like
“I wonder what I can do without being encumbered by X”. Then everyone else jumping up and considering it a replacement instead of a research project (I imagine the excitement is due to people being fed up with all the craftiness found in X).
For one thing the Kernel is a lot younger than X. It also has far more developers and a structure that allows it to stay in continual development. X has very few developers and it’s architecture frightens a lot of people away (there isn’t much influx of new developers). If a more maintainable modern display server is used then chances are that new features are added more quickly and more developers join, two things that are the biggest weakness in X.