Wayland Preparing For 1.0 Stable Release

[kraftman]

Totally agree.

To all of the network-transpara-tards: Network transparency in X was somebody's great idea before people had properly figured out the right way to do that kind of thing: web applications, and/or VNC. You might as well get on board with doing things the way they should be done, or commit yourself to using old computers, old operating systems, and old applications. Have fun with that.

Just to note I was talking about non Linux operating systems. There are many cool features in Linux that aren't used because some old crap is years behind. I see no reason to care about bsd and it will be best if developers simply focus on Linux and ignore others.

[Scias]

I don't quite understand what you mean with X offering horrible desktop performance and how Wayland will increase such performance. I've had no problems running all sorts of eyecandy on my desktop under X, and using programs like Blender/Maya/Mudbox certainly hasn't been slower than under Windows. If Wayland increases performance then great (I've always thought performance was a driver issue, not bound to X) but it's not as I've ever found X server to be some kind of bottleneck. Am I missing something?

I was talking about the 2D windowing part, not about the 3D. 3D programs run fine if not faster under Linux (with decent drivers of course). The 3D apps bypass X anyways.

But did you ever try to resize a heavy window under Windows or OSX, and did the same in Linux under a composited environment ? Or ran a 3D game under the same conditions ? You will notice how sluggish it is on Linux compared to the other mentioned platforms. It's funny because Compositing is supposed to reduce overhead and improve desktop performance, but on X11 it is and never has been the case, first because of broken drivers but it wouldn't be surprising if the overcomplex architecture of X -that by the way, was not designed for compositing or hardware acceleration tasks in the first place- wasn't one of the possible causes of the slow desktop performance.

The fact that Wayland is way simpler/lighter and is already designed for all this latest stuff might improve things quite a bit don't you think?

[phred14]

Obviously, the Wayland folks aren't going to overlook the ability to graphically access a machine remotely, since VNC/RDP/etc... are so popular in business environments, but FFS, can we please abandon the idea that a window manager should have network transparency? It is 110% preferable to just stream the whole desktop ala VNC, LTSP, or maybe even a brand new protocol that's more efficient, rather than using all of these nifty little X11 over SSH tricks. Wayland will also probably be an excellent first step towards improving the performance of remote desktops on Linux, which still lag way behind the performance and smoothness of Windows RDP.

Wayland addresses a real need for 99% of users, and that is replacing the awful performance, esoteric API, and legacy bloat of X11, with something modern that's designed for the way people use computers now. People who need legacy features of X11 should just commit to doing things in a way that will actually be supported, or just resolve themselves to using old computers running old OS to do things the same way they were done in the 90s. Meanwhile all of the young whippersnappers will use that newfangled Wayland thing on their shiny new computers.

There's the implicit assumption here that the remote machine -has- a desktop. Of the 4 cases I mentioned for 3 of them there is no remote desktop. In fact my servers at home don't even have an X server installed, let alone desktop. They have enough of X installed to run X clients. The only case where I'm connecting to a desktop is where I'm connecting from home to my already-running laptop using x11vnc/vncviewer.

You talk about X being bloat. I'll cover that and up you one - for my purposes the desktop is often bloat. Also in my case it sometimes really matters. For VLSI CAD sometimes I'm pushing these machines right to their limits, and that means running as little on the machine as I can get away with, just to make more room for my "real work".

Notice again, I haven't said anything "anti-Wayland" here. I'm just outline my very real needs, which I don't think are unique - just "industrial".

I'd be much happier about this whole discussion if someone on the pro-Wayland side said, "Here's our roadmap, and here's how we're going to cover needs like yours." Instead I'm being called a "tard" and told to either go away or get modern. While I may also resent the personal slur to that, I'll also say that I'm also one of the most progressive Linux users in the F500 company where I work. When you want to "move Linux" you need to realize that in spite of its miniscule 1.5% desktop share it has an incredible baked-in presence in industry. There is an enormous amount of legacy to keep working, and perhaps X-over-Wayland is the right way to handle that. But other than hand-waving by basically sensible people here, usually countered by lose-the-past arguments by others, I haven't seen much. I'm about the last one to go in for foil-ware and excess planning, but I'd really feel better with a roadmap.

Acceptance of Wayland would go a long way if they'd just publish a roadmap showing how they handle trivial things like legacy industrial/corporate use.

[phred14]

Wayland addresses a real need for 99% of users, and that is replacing the awful performance, esoteric API, and legacy bloat of X11, with something modern that's designed for the way people use computers now. People who need legacy features of X11 should just commit to doing things in a way that will actually be supported, or just resolve themselves to using old computers running old OS to do things the same way they were done in the 90s. Meanwhile all of the young whippersnappers will use that newfangled Wayland thing on their shiny new computers.

99% of all users? Who do you think are Linux users?

I'll guess that I'm in a bigger camp of Linux users than you are. Have you considered that for instance, most VLSI CAD (what I do) worldwide has moved to Linux, most film production and special effects have moved to Linux, heavy-duty numeric simulation such as weather forecasting and nuclear simulation, etc, etc, etc. There's a LOT of industrial Linux out here, and if you think I'm a "tard", I'm one of the most progressive Linux users in my F500 company.

The last time I saw this attitude, I was trying to pick up a new programming language, and asking about facilities for interfacing with legacy binary data. I was told that I should "get with the program", rewrite all of that old stuff in the new language, and eliminate the need for interfacing with old data. Yeah, right.

There's an enormous amount of legacy out there. The way to make progress is to make sure that legacy can be handled in some way or another, if only layered on top of your shiny new solution. Ignoring the legacy or denouncing it as a dinosaur will simply guarantee that your shiny new solution will never get widespread adoption.

[Ole Laursen]

phred14: I think the reason there's no roadmap beyond "X11 will still work" is simply that they aren't there yet at all.

Looking at the status pages, it looks like the toolkits still have a bunch of things that aren't handled, and besides, without real usage from a desktop environment like GNOME or KDE, there are probably still some nasty things lurking they need to think about and solve. I personally think it's unrealistic we'll see a shippable native GNOME 3 on Wayland this year.

I'm using X11 apps over the network every day myself too, but I'm actually glad they are waiting with the networked aspects until they have something that works for the local case. We don't need something half-baked.

In the meantime, X11 will be working just fine (both with and without Wayland), so it's not that we're any worse off. Despite the ubiquity of X11, it does suck horribly over high-latency links. Maybe a new network layer will actually be able to address this (I know NX, and it's much better, but was just too buggy for me).

[phred14]

phred14: I think the reason there's no roadmap beyond "X11 will still work" is simply that they aren't there yet at all.
...
I'm using X11 apps over the network every day myself too, but I'm actually glad they are waiting with the networked aspects until they have something that works for the local case. We don't need something half-baked.
...

A roadmap doesn't have to say, "Here's what we're going to do," it just needs to say, "This is one of our objectives." and better yet, "We're going to start looking at it at approximately this date." I can agree that they shouldn't be in too big a hurry to start adding network transparency. My distress is that every time the issue comes up, most of the response is, "You're a tard for wanting that!" A much better response would be, "We will start working on a network-transparent solution after we have the basics in place."

[89c51]

My distress is that every time the issue comes up, most of the response is, "You're a tard for wanting that!" A much better response would be, "We will start working on a network-transparent solution after we have the basics in place."

But you will get some form of network infrastructure. And if i understand elanthis's post well it can be done more than just one way. Instead of having it in the protocol as X they will implement it on another level. And maybe you will have more solutions than one.

And yes Wayland will not replace X in every setup immediately. It will first be tested by normal/average users (Ubuntu Fedora etc) and then will find its way in the Professional field (graphics CAD engineering etc) while keeping compatibility with X for all the apps that still rely on X.

[Silverthorn]

A roadmap doesn't have to say, "Here's what we're going to do," it just needs to say, "This is one of our objectives." and better yet, "We're going to start looking at it at approximately this date." I can agree that they shouldn't be in too big a hurry to start adding network transparency. My distress is that every time the issue comes up, most of the response is, "You're a tard for wanting that!" A much better response would be, "We will start working on a network-transparent solution after we have the basics in place."

The solution could be very simple.

Most applications are written in Gtk or Qt.


If a HTML backend is used (for example Gtk broadway), this could be the future usage scenario:

1. The user connects to the remote host by ssh.
2. The user request that the applications should be drawn locally on the host.
3. The host redirects the host application to use the HTML backend.
4. The resulting web page is transferred to the guest.
5. A HTML browser is used to view the final results on the host.

The only missing piece is the lack of support when ssh is used. In fact, what you really want is for the ssh client/server to be updated to handle different protocols other than the ones provided through X. And maybe you could even start using Gtk application right now through ssh port forwarding??

[leeenux]

99% of all users? Who do you think are Linux users?

I'll guess that I'm in a bigger camp of Linux users than you are. Have you considered that for instance, most VLSI CAD (what I do) worldwide has moved to Linux, most film production and special effects have moved to Linux, heavy-duty numeric simulation such as weather forecasting and nuclear simulation, etc, etc, etc. There's a LOT of industrial Linux out here, and if you think I'm a "tard", I'm one of the most progressive Linux users in my F500 company.

The last time I saw this attitude, I was trying to pick up a new programming language, and asking about facilities for interfacing with legacy binary data. I was told that I should "get with the program", rewrite all of that old stuff in the new language, and eliminate the need for interfacing with old data. Yeah, right.

There's an enormous amount of legacy out there. The way to make progress is to make sure that legacy can be handled in some way or another, if only layered on top of your shiny new solution. Ignoring the legacy or denouncing it as a dinosaur will simply guarantee that your shiny new solution will never get widespread adoption.

I'm guessing you're about 100 years old, and very stuck in your ways. The Linux community has no obligation to continue supporting such a crusty old standard, and all of the big names have made it abundantly clear that they'll jump ship to Wayland the minute it's ready. This is the evolution of FOSS, whereas Windows has an obligation to keep supporting their old cruft, because there's too much other proprietary legacy cruft that depends on it. FOSS can get forked if needed and change with the times, proprietary business applications can't always do that.

I'm not even sure why you're trying to convince me, my thoughts on Wayland aren't even important. I'm just a dual Windows/Linux Systems Engineer with a long, illustrious resume filled with uber-scale enterprise IT at Fortune 500 companies (all the way up to Fortune 10: eat your heart out), who also happens to own a couple of marginally popular, very niche, and still relatively-insignificant-in-the-greater-scheme-of-things open source projects. You can whine to me all day long about how nobody should expect you to ever have to change the way you do things, but at the end of the day, I won't be the one who takes X11 away from you.

Your efforts would be better spent lobbying the Wayland folks to be more like X11 and the distro owners not to adopt Wayland. They'll love to hear how you know better than they do, because you work at a Fortune 500 company. After all, Fortune 500 companies only employ about 100 million people worldwide to work in IT or related engineering positions like yours. Out of all of them, however, you're the only one who repeatedly mentions it in his posts like it's unique or special, and like half of the people in a Linux forum don't also work at Fortune 500 companies. You work at a Fortune 500 company, I get it. Well done. Have a cookie.

[leeenux]

The only missing piece is the lack of support when ssh is used. In fact, what you really want is for the ssh client/server to be updated to handle different protocols other than the ones provided through X. And maybe you could even start using Gtk application right now through ssh port forwarding??

The thing is, what Linux application or admin task can't be done strictly in SSH from the command line? I would say that running a headless server for remoting GUI applications is a pretty odd configuration... Either install a full DE and run VNC or LTSP for GUI apps, or do it all from the command line like Linux admins have been doing since the dawn of time. Linux will never evolve if people keep catering to geezers like Fred.

A list of common Linux tasks, that CAN be done CLI style:

FTPing/SCPing/RSYNCing files
Configuring Apache/BIND/CA/Samba/iSCSI-Target/OpenLDAP/pretty-much-anything-else
Everything that's done on super computers, 'cause you know they ain't running Gnome on them
MySQL/PostgreSQL/your-favorite-NoSQL

Things that CAN'T be done CLI style:

Apparently CAD, according to phred
...and that's about it.