Ubuntu Allegedly To Have Its Own X, Wayland Alternative

[johnc]

I recently evaluated all of the other distros since Ubuntu 12.10 was such a turd.

What I discovered was that even though 12.10 sucks, it was still better than anybody else's.

Well this is where I feel like I'm up crap's creek. It seems that each new release of Ubuntu is worse than the prior and Canonical seems completely pleased knowing the turd is swirling around the bowl. But as bad as it is, the only reasonable alternative seems to be to abandon Linux altogether. I think I'll just stick with 12.04 for as long as possible, and try to work around its shortcomings.

So Android comes out of nowhere with it's own display server to become the dominant consumer OS, better than Windows OSX, iOS or anything else. Maybe Wayland isn't as good as everybody claims, and if Ubuntu can write something that's small, simple, and fast, and STILL beat Wayland to market, then why shouldn't they? Should they have also just went with Gnome3 instead of Unity so that they could be as crappy as Fedora 18? Or why not the perpetually buggy KDE? In all fairness though, the latest Gnome3 is so buggy that KDE is probably the stable one now, and I never thought I'd say that. More competition in Linux standards is great, because a some of the current standards suck and need to be dethroned by a new contender.

Those of you that think Ubuntu should wait until Ubuntu 14.10, 16.10, or whenever to adopt Wayland when it's finally ready, probably have no credentials to be telling Ubuntu how to operate. Ubuntu just works, and the only other distro that can make that claim is RHEL and it's copy-cats, and RHEL accomplishes that by sitting on really old versions of software. Ubuntu is the only distro in LInux history that can properly do bleeding edge kernels and other software, I trust them over the armchair-quarterbacks of the internet.

I've never been into the Wayland hype myself, as the project claims to bring pretty much nothing better than X except cleaner code and protocols. From a user perspective, big whoop. As far as I can see, no real new "features" are actually planned, and Windows 7 & 8 continue to have far more advanced display servers.

So in that sense, if Canonical wanted to do something better I could definitely get excited. But it's Canonical. And looking at the way things have been going recently, nearly every move is hair-brained and hardly beneficial for the user. And, worse, quality is garbage. If Apple's slogan was "It just works!" then Canonical's is clearly "It just doesn't work... and we don't care... and we're pushing it out the door anyway." So I have very high doubts about their ability to pull off a desktop-quality display server without cutting major corners. In 12.10 the end-user could not even *install* proprietary drivers without borking his system. And this group is going to write a whole display server that works?

Canonical keeps talking about bringing in the new users, and getting the non-tech savvy people into the fold. You know... "Linux for human beings." But they seem to be avoiding the fact that critical show-stopping bugs (like black screens on boot -- insanely common in Ubuntu now) are absolutely unacceptable to non-tech savvy people. Yeah maybe we can dump into a console and restart LightDM or re-install the nvidia-current package or whatever the solution, but expecting that of people who are new to Linux is just asinine.

And they don't care. They're more concerned with all these superficial, valueless changes like collapsing the Software Updater or having the Scopes perform searches on a central database or whatever other "improvement" that is entirely moot when all I've got is a blinking cursor.

[Ericg]

I recently evaluated all of the other distros since Ubuntu 12.10 was such a turd.

What I discovered was that even though 12.10 sucks, it was still better than anybody else's.

1. Mint - slightly less stable than Ubuntu, less useful desktop, it has nothing on Ubuntu
2. Arch - A toy, not a real distro meant for real work
3. Fedora - 17 was decent(except for the bugs), 18 is unbelievably shitty
4. OpenSuse - Are they still around?
5. Mageia - Might be contender someday, but not yet
6. Debian - I appreciate them being the extremely FOSS-minded, but they always have a dinosaur kernel that can't boot a newer PC. Debian Unstable is... Unstable.

1) Less stable, better desktop
3) Agreed
4) Surprisingly, yes
5) So far theyre doing good, hope to see them continue
6) Agree on both points

2) Very much so disagree... All computers I've got run Arch both as desktops and 2 dedicated home servers. Its hardly a toy. Is it for the average user? Hell no, but it is far from a toy. Once you get it setup you just stop caring, keep up with updates, check the homepage for news once a month to spot any update-problems or non-standard updates (replacements for example) and you just keep going. Btw my "real work" is CS, schooling, video/audio editing and casual gaming-- its really nice having the latest versions of all libraries, LibreOffice, AV tools and gfx drivers & wine to get those things done.

[Ericg]

I've never been into the Wayland hype myself, as the project claims to bring pretty much nothing better than X except cleaner code and protocols. From a user perspective, big whoop. As far as I can see, no real new "features" are actually planned, and Windows 7 & 8 continue to have far more advanced display servers.

1) Cleaner code,
2) cleaner and more efficient protocol,
3) better support for GPU-switching,
4) better support for full-screen apps (try having a full screen window change the resolution under X, depending on your WM you'll get mixed results)
5) proper screenlocking (under X an app can bypass the screenlock if it really wants to)
6) lightweight enough that it can be used on tablets, phones or desktops.
7) Better multi-monitor support,
8) frame-perfect graphics-- no lagging, tearing or stuttering
9) also fun fact: its relatively future-proof. The wayland protocol really just moves pointers and buffers around to different sections, it lets other programs handle what those buffers contain, how the data got there and things like that. By letting third party handle the major implementation details and just having the protocol as barebones as possible it means if there ever comes a day when we have a big shift in how we do graphics (like X went through) we may not have to throw the protocol out the window or start bypassing it like we did with X, we just change the 3rd party bits to the new architecture./ style and keep plugging away

Any questions?

[Akka]

Yes Arch is fantastic.

• It decently stable, more or less as stable as the 2 releases a year desktops
• You always has the latest libs.
• No messy big updates two time a year. Only practical smaller updates continuity.
• Good documentation
• Everything works as it's supposed, a minimum of silly patches of upstream project.

[Figueiredo]

Any questions?

I have one: What are the key differences between wayland and android's DS?

[Akka]

1)
5) proper screenlocking (under X an app can bypass the screenlock if it really wants to)

Shouldn't the new sceenlocking in the composting layer in kde and gnome do the same thing?

[Ericg]

I have one: What are the key differences between wayland and android's DS?

Literally the only reference I can find online, because I dont know that one off the top of my head, is: elinux.org/images/7/79/Lf_elc12_rosenkranzer.pdf

In it (slide 19 and 20) they mention its "Designed for modern GPU's" "accessible from C, Java and C++" with "little overhead" BUT just going off of experience and anecdotal evidence, its a battery-killer (problem for laptops if you wanted it on "desktop" linux), judging how Google handled the audio portion of Android, its probably not well written or they took a lot of shortcuts. And I'm gonna guess it doesn't scale well for the simple fact it was designed with phones in mind wherein you either have the "Desktop" up (home screens) or 1 fullscreen window. Not 2 or 3 windows with one rendering video the other doing CAD work and the other writing a memo.

Obviously Im exaggerating with that example but you get my point-- its designed for 1 app at a time, not multiple all rendering at once. Also it was designed around touchscreens which may or may not be an issue, you'd have to go through all the code and see how hard it would be to allow for mouse style events.

[Ericg]

Shouldn't the new sceenlocking in the composting layer in kde and gnome do the same thing?

Ideal-world yes but then we're back to the core issue of wayland: "Why didn't you just extend the X server?" ---> "Because most of our time is spent bypassing the X server anyway. Lets just get it out of the way so we can finally do the things with graphics that we've always wanted." Also it doesn't help if composting in KDE is turned off, which it sometimes is. Gnome (3) users dont have a choice in that regard.

KDE and Gnome screenlocking also doesn't help if the user isnt USING those two screenlockers. Any of the *Boxes I believe still use the default X screenlocker, with Wayland every compositor will have proper screenlockin at a minimum and if KDE, Gnome, Qt, etc want to do something fancy ontop of it, thats fine.

[Figueiredo]

Literally the only reference I can find online, because I dont know that one off the top of my head, is: elinux.org/images/7/79/Lf_elc12_rosenkranzer.pdf

In it (slide 19 and 20) they mention its "Designed for modern GPU's" "accessible from C, Java and C++" with "little overhead" BUT just going off of experience and anecdotal evidence, its a battery-killer (problem for laptops if you wanted it on "desktop" linux), judging how Google handled the audio portion of Android, its probably not well written or they took a lot of shortcuts. And I'm gonna guess it doesn't scale well for the simple fact it was designed with phones in mind wherein you either have the "Desktop" up (home screens) or 1 fullscreen window. Not 2 or 3 windows with one rendering video the other doing CAD work and the other writing a memo.

Obviously Im exaggerating with that example but you get my point-- its designed for 1 app at a time, not multiple all rendering at once. Also it was designed around touchscreens which may or may not be an issue, you'd have to go through all the code and see how hard it would be to allow for mouse style events.

There are android phones that can output a different screen than that of the phone. My moto atrix from the top of my head, so may not be impossible.

But from your answer another question arises, can canonical extend android's DS enough that it would also work well in the desktop form factor?

My point is: Canonical is faced with a choice: They can follow intel and the other GNU distros and adopt wayland, relying on Intel and community to further develop it, also requiring porting of the desktop GPU blobs to this DS supported only by GNU distros and also of the apps/toolkits. Or, they can follow google and adopt android/chromeOS DS and instantly gain support for ARM SoCs drivers, which would require convincing AMD and nvidia to port their blobs to android/chrome OS, the porting of the apps and toolkits and rely on google to further develop the DS.

Neither is painless, but I can understand canonical if they choose to stand closer to a google/linux instead of GNU/linux. Google has already taken linux where no other GNU/linux distro has ever gone: consumer devices. By "borrowing" hardware support from android/chromeOS, ubuntu can deliver a better user experience and has a higher chance of convincing an OEM to ship a product with it.

Maybe wayland is technically better, but that is meaningless if it only performs well in a very strict number of devices.

[Ericg]

There are android phones that can output a different screen than that of the phone. My moto atrix from the top of my head, so may not be impossible.

But from your answer another question arises, can canonical extend android's DS enough that it would also work well in the desktop form factor?

My point is: Canonical is faced with a choice: They can follow intel and the other GNU distros and adopt wayland, relying on Intel and community to further develop it, also requiring porting of the desktop GPU blobs to this DS supported only by GNU distros and also of the apps/toolkits. Or, they can follow google and adopt android/chromeOS DS and instantly gain support for ARM SoCs drivers, which would require convincing AMD and nvidia to port their blobs to android/chrome OS, the porting of the apps and toolkits and rely on google to further develop the DS.

Neither is painless, but I can understand canonical if they choose to stand closer to a google/linux instead of GNU/linux. Google has already taken linux where no other GNU/linux distro has ever gone: consumer devices. By "borrowing" hardware support from android/chromeOS, ubuntu can deliver a better user experience and has a higher chance of convincing an OEM to ship a product with it.

Maybe wayland is technically better, but that is meaningless if it only performs well in a very strict number of devices.

Well Wayland's already backed by a few car manufacturers and a few mobile manufacturers, its got Intel's, Red Hat's and all of X.org's backing, 90% of the community and all major toolkits. If Canonical really wants to get into bed with google thats their choice...but I think it'd kill Ubuntu.