Slashdot Mirror
GNOME 3.10 Is Now Properly Supported On Wayland
An anonymous reader writes "One week ahead of the GNOME 3.10 release, all of the basic Wayland support for GNOME has been merged. With today's GNOME Shell 3.9.92 release the Wayland branch was merged and there was also an updated Mutter Wayland release, besides earlier GNOME 3.9.x packages fostering the Wayland support. Fedora 20 is expected to ship with GNOME on Wayland as a technology preview. Additional details about the current GNOME Wayland support are available from the GNOME Wiki."
(and didn't want to google it):
http://en.wikipedia.org/wiki/Wayland_%28display_server_protocol%29
http://wayland.freedesktop.org/
Wayland
Wayland is intended as a simpler replacement for X, easier to develop and maintain.
Now I can complain about the user interface on a whole new display technology!
No doubt, I am cheering the open source drivers to continue their great progress but I can't understand why Nvidia and AMD don't enable EGL extensions on their desktop drivers (especially AMD since I'm a shareholder because they started supporting open source). With Mir and Wayland needing the extensions, Gabe Newell saying Linux is the future of gaming, and the future of Linux windowing being Mir or Wayland, I'm not going to get super excited until one of the Big Two GPU vendors start supporting it.
And I'm hoping it's you, AMD, that will be the first to claim that crown on Linux. Please let it be in the forthcoming hardware Newell mentioned.
Really disappointing choice of version numbers.
Repeal the 17th Amendment TODAY! Also Please Read http://www.gnu.org/philosophy/right-to-read.html
More importantly, when will we see GNOME 2 support?
But in fact it's NOT properly supported.
If you read the wiki that the article poster linked to, there are all sorts of caveats and missing functionality. "Properly Supported" means functional parity, and from where it sits right now, there is not functional parity.
This affects non Gnome 3 users sometimes (e.g. File/Open puts you into "Recently Used", wasting a bit of your time and clicks, in a app that uses GTK3.)
Ooohh, is that what it is. Is there a workaround?
Here's one that keeps the "Recently Used" category empty. Unfortunately, it does not prevent GTK3 applications from defaulting to that absurd category in a File/Open operation. As a logged-in user, run:
rm ~/.local/share/recently-used.xbel
mkdir -p ~/.config/gtk-3.0
echo -e "[Settings]\ngtk-recent-files-max-age=0\ngtk-recent-files-limit=0" > ~/.config/gtk-3.0/settings.ini
rm ~/.local/share/recently-used.xbel
The second rm will probably cause an error message, unless some application is busily updating the "Recently Used" category while you run these commands.
Those who can make you believe absurdities can make you commit atrocities. - Voltaire
It's not that Wayland does a whole bunch that X doesn't, it's that X has a lot more hoops to jump through to keep going. Wayland just presents what amounts of a framebuffer and a simple protocol to let the compositor and clients communicate about size changes, movement, available displays, etc.
All of the modern graphical environments and applications are using the COMPOSITE extension to X, which adds an extra step to a lot of graphical operations. Plus, to be "X", you have to support things like the old X line-drawing primitives, fonts in the server, and other things that simply aren't used anymore. Important things like changing the screen resolution are kept in protocol extensions that you have to check for before using. Large amounts of code and protocol are dedicated to working with screens of vastly different capabilities - everything from monochrome monitors to "true color" displays. Nobody has a fixed-sized monochrome X terminal anymore, but the code has to account for it still.
Plus X stores a ton of things in the server, making it big, slow, and a source of potential security/information disclosure issues. Wayland stores less and does less.
In other words, developers are hamstrung having to maintain and work around lots and lots of very old code that will never, ever be used by a new application, ever, but has to be there, even though it slows things down, takes up space, and makes things more complicated.
Personally, I would've liked to have seen something more like "make COMPOSITE a part of the core X protocol and deprecate lots of things" and see X slowly evolve into a more "modern" system, but that's just me.
As for GNOME - I realize that GNOME 3 is different from GNOME 2, but I'm at least happy that for once the Open Source community *tried something different* instead of just aping Windows or Mac OS X (though GNOME 3 is obviously inspired by the latter). Maybe it worked, maybe it didn't, but at least we can claim to attempt to lead, instead of just blindly following.
...but it's being eaten...by some...Linux or something...
This is a great leap forward for desktop Linux and we must remember the open source luminaries that have made this advance possible, starting with Mr. Mark Shuttleworth. Mark committed to making significant contributions to Wayland back in 2010, and generously offered to support KDE and Gnome in the transition. Wait, what? They never contributed a single line of code? They were secretly working on another project and are now in a pissing match with Intel??
Check this presentation by Daniel Stone (one of the X.org developers) on the problems with X.~
http://www.youtube.com/watch?v=RIctzAQOe44
Have they fixed the lack of options for decent previews? Selecting a picture or video by a 20px preview just doesn't cut it anymore.
Yes, it's called using KDE.
--
BMO
Anybody who has looked in the innards of X knows its a pig. No secret there. It's only Unix fanbois that cannot fathom that some parts of Unix were not properly designed from inception.
At 15:19, David Stone has a nice slide that says:
xserver 1.0.2: 879,403 lines of code
xserver (now): 562,678 lines of code
That's 300,000+ lines of cruft they wiped out without breaking the X protocol. Wayland is currently about 20,000 lines of code, that's about 3.5% the size. Even if that doubles they're still getting rid of 90%+ of the old code, that's huge.
Live today, because you never know what tomorrow brings
They couldn't fix X. The developers behind wayland looked at what it would take to do the same thing they are doing but do it with updates to the X code. They saw a process that would take 20 years rather than 5.
X has 20+ years of legacy code, on a modern X install you are using at best 5% of the code. When you've reached a point where 95% of the code and application isn't even being used it's time to start over rather than work to upgrade what you have. You would spend so much time troubleshooting old code that you could literally rewrite the entire thing 4 times before you finish.
Wayland is designed like a modern interface that is used on almost every other OS. It's designed to be as simple as possible while being extensible. It isn't trying to be anything but a channel for programs to talk to graphics hardware. X was everything and the kitchensink as you said everything from drawing lines on vector displays to keyboard and mouse. X was a monster that's time to deprecate and replace.
X was so "ahead of its time" that its entire architecture was dumped in version 10 to give way to X11, and then it remained so far ahead of its time that to this day NextOS, MacOS, Android and Windows have yet to adopt a single thing from it, contrary to the rest of Unix most of which has made its way into those operating systems.
And no, it was not designed to access resources from the desktop. It was mainly designed so that you could use a dumb terminal to access your server. When it became clear that was pie on the sky, instead of redesigning the turd, they just added layer upon layer of cruft, so you ended up with a dumb as doornails protocol running on a heavy weight, expensive "dumb" terminal.
Wayland is an effort to remove those layers of cruft that nobody uses (Xtoolkit?)
Lastly the web browser has nothing to do with Unix. It is platform independent. The fact that you think the web==unix shows how little you know about deep OS architecture.
The dumb terminal at that time was a VT100.
By dumb terminal I mean a thin client, something that didn't happen.
X mostly ran on engineering workstations.
Not by design dude. It was meant to run on thin clients but it ended up being such a pig that you needed a workstation to run it. At the time a "thin" X-client was more expensive that a PC.
I'm not sure why you think that is "pie in the sky" since it worked and continues to work rather well.
So a protocol designed to run on a cheap thin client ends up needing a powerful workstation to run and you call it "working rather well"? With those definitions it's no wonder you consider X a success.
Part of the reason for that was because the protocol was rich enough to transmit graphics primitives at a higher level than a bitmap. Nothing dumb about it.
No one uses such primitives because they are incredibly sucky. Hence VNC and such.
I did find NextStep and NeWS superior to X11 and it's a damn shame they didn't succeed
Of course, anything was better than X11. Steve Jobs famously declared "X11 is brain dead".