Slashdot Mirror
First Steps Towards Network Transparency For Wayland (phoronix.com)
munwin99 writes: For the longest time, when bringing up Wayland a recurring question was 'what about network transparency?!' Well, Samsung's Derek Foreman has today published the set of Wayland patches for providing Wayland network transparency by pushing the Wayland protocol over TCP/IP.
What people want is ssh -X and yes it is a top priority to many.
And yet, every time the subject of Wayland comes up, you can expect that at least 25% of the posts here will be concerning network transparency or their responses.
That's why it wasn't considered when the protocol has been designed (and rightfully so), but it's great to see it as a later addition.
I have never used network transparency in X for any significant purpose, but it was great for quick hacks, especially when I had a smartphone running X.
Holy fuck, how about they actually make it simple to run Wayland?!
I mean they've been working on Wayland for years now, yet it's still a real pain in the ass to get working on a modern Linux distro.
As shitty as X.org is, at least it's fairly easy to install and get working these days. It usually just happens as part of the Linux distro installation.
But getting Wayland running? Holy fucking moley! Be prepared for a fight!
The best I've managed so far was getting some Wayland-in-X thing running, and the results were less than spectacular.
I don't give a fuck about its support for network transparency when I can't even get the fucker to run on my systems!
They should at least get it to the point where it can be used on a standalone workstation, and only then should they look into network transparency.
A windowing system that we can't actually use is, well, pretty fucking useless!
Personally, I would never use any windowing system by choice that did *not* have network transparency. Non-local VMs and applications with specific hardware requirements or physical attachments are the biggest (as specific examples that I have used *today*). I use VNC heavily (including KVM-to-VNC for boot level interactions with systems) but that is no overall solution since it doesn't give you integrated desktops usually (copy-paste, breaking out each remote window into a local window, etc).
I'd certainly accept something like (I currently use it) NX (No-Machine's X) when run in rootless mode. That works decently well for allowing remote GUIs to behave more or less like they were local.
Previously, the developers always refused to consider network transparency, and heated discussions followed. If now it is accepted, it is newsworthy for those who care about the feature, even though nobody can actually run Wayland yet.
Those who do not understand X are condemned to reinvent it, poorly.
-- Harry Spencer (sort of).
I concur. VMs, embedded hardware, headless machines- I'm on them all day. And ssh -X is all that I need working for my environment. As long as that works, everything else just is seamless. I think we're not going to see a reduction in VM's. And the number/amount of embedded hardware's only growing.
Now, X certainly has ugly warts. I'm hopeful for what Wayland's offering. This network transparency patch for Wayland sounds like a great start.
--Mark
Wake me up in several or more years when something is actually available, works, and is really backwards compatible. Meanwhile, those of us who depend on thin clients really do have a problem with throwing away X11.
You do not use a full session, but often need to launch small applications without disturbing the normal use of the remote machine. For example it could be the software manager of the distro, or a specialized software only available on a particular machine (for example because it has a hardlock key, or it does not run on the operating system you are using on you main workstation).
I do, why am I wrong?
Typical usage : I log onto distant machine, start working in command line (vim, python, matlab -nodesktop), then at some point I will need to display a couple of graphs or images. That's a relatively small graphical payload for which I *do not* want to use VNC. With ssh -X I get the windows to be displayed locally just as if I was doing the work on my light-weight terminal.
Lets analyse how long this time is. The initial wayland release was on 09 February 2012...
The 'what about network transparency?!' concern was appearing long before the initial release. Here is the usual huge thread about it from 2008: http://linux.slashdot.org/comments.pl?sid=1017147&cid=25619591
About 7 months ago I began using Linux on a headless workstation via VNC as my primary interface for development work. On my gigabit LAN the performance is amazing; certainly at least as good as remote X, but without all the font and window manager glitches inherent to remote X, and I don't need an X server on my desktop. The server software is TightVNC and the viewer is TigerVNC. It's actually far better than RDP from a new Windows 10 laptop on the same network. All Wayland would have to do is match that and I'm good with it. At least on fast local network.
Maw! Fire up the karma burner!
ssh -X is seriously is the slowest shit. Barely usable on a 100 megabit LAN even with compression enabled.
Your basement stuffed full of old Pentium 3 boxes does not count.
That's my living room!
It just LOOKS like a basement.
You have the right to remain sentient. If you give up the right to remain sentient, you will be elected to public office
Wayland is a fairly controversial replacement for X11, written by the people currently maintaining the X.org X11 stack.
As the summary implies, Wayland been criticized for lacking significant features of X11 such as network transparency. Defenders have argued that network transparency is a minority application and that they don't like the way it's implemented in X11 anyway,
Those of us who use network transparency are rather bothered by being told that something that works fine for us (and it does, I regularly have to configure LibreOffice systems running on AWS instances, and have never bumped into any of the supposed problems Wayland advocates insist I have) are things we don't really need or want. We're not happy about losing functionality simply so that someone can go from 59fps to 59.5fps when playing Call of Duty.
Previous proposals have varied from proposals for an optional intermediary protocol sitting between Wayland and the client (apparently by people who have no idea what the transparency part of "Network transparency") and even the ability to stream the contents of Windows using H.264.
This proposal sounds, at least at first glance, to be better than those hacks. Hopefully it means they're finally taking the issue seriously.
You are not alone. This is not normal. None of this is normal.
It's mostly in reference to "top priority" and "majority". There are some niche use cases of course, but in general it's only useful for some random ad hoc stuff.
-SR
I haven't found a better way to persistently run remote graphical programs than with X11 via xpra.
What do you suggest instead? VNC? RDP?
"Nine times out of ten, starting a fire is not the best way to solve the problem." - my wife
This is ludicrous. Wayland is hardly "languishing", and there has been no push on anyone's part to make it the "de factor Linux GUI". It is under development, everyone recognizes it as the next way forward, toolkits and drivers are targeting and supporting it, and it will be adopted when it is ready for adoption.
Anyone using the interim period to scream about network transparency is a moron. Yes, that means you.
RDP indeed is a very good protocol. It's very fast, much faster than X11 forwarding, and can forward files, printers, and sounds across the link. Typically it's faster than VNC too. At one time there was talk about making a wayland module that would use RDP as the underlying protocol to remote Wayland windows and applications across the network. This actually makes more sense to me than forwarding the wayland protocol itself. RDP can do per-app forwarding (like we're used to on Linux), or the full desktop.
Are you serious? It's 2016 and the rage is cloud computing with distributed virtual machines and containers all running programs. You better believe remoting and network transparency is in demand, and actually essential. Apps could be local in a docker container or on the cloud. All interfaced on a laptop or tablet together seamlessly. Really it's the old 1990s Sun vision actually materializing.
You have never administered a really stripped-down server, have you?
If you think that you have, then you don't know the meaning of "really stripped-down".
The real "Libtards" are the Libertarians!
Actually the summary doesn't mention X11 at all. Even that would have helped. It just says that Wayland has been criticized for lacking network transparency and a set of patches were released for pushing the protocol over TCP/IP. I had no idea what Wayland was from the summary because I've been away from the Linux movement for a while (I used to be system administrator). Even just just including the word X11 in the summary would have made it much clearer.
LTSP
It is pitch black. You are likely to be eaten by a grue.
Seriously, if they ever want to make it the year of Linux on the desktop they would adopt RDP as the protocol (as in compatible with mstsc.exe). It'd be a massive potential userbase of people running windows who could be immediate users of free software. I could see virtual desktops as a mass-market business, not the niche corporate Citrix/TS/VDI thing it is now.
I always wonder why I don't make my own dekstop a VM and quit customizing or even caring if the actual machines I connect from do anything other than run the remote access client well.
What "really stripped-down server" were you working on that you needed to launch graphical applications to administer it?
A "really stripped-down server" is, to me, a server without vi or nano.
Yes, that "random ad hoc" stuff is called the capabilities of the computer environment. Kind of like reading CDs and USB keys. Most people most of the time don't do it, but I don't think you want to pull those bits out of the kernel. Other than a tiny minority of technical users, most people don't write C code either. Should we drop the C compiler from the standard Linux environment too? Why not go whole horse and just use Windows? No compiler, no remote terminal, no native nothing without third-party add-ons. Just because most of your friends aren't capable of taking the full advantage of a Linux machine with x11 doesn't mean you get to piss all over it and fuck it up for the rest of us.
There are so many reasons why VNC is actually a *superior* way to run graphical software remotely that it isnt even worth discussing.
You, sir, win at being a fool. VNC is a horrid abomination. It sends bitmaps of changed screen areas. What is possibly superior about that?
X sends drawing primitives (draw rectangle here, draw button there, etc). Assuming the remote end is able to render those primitives it can actually be a very efficient protocol. Now, most L-users want the shiny and bitmap-rendered gradients with OpenGL transparency, etc. That is difficult to send as a series of primitives. Remote OpenGL support kind of works if you can run an X server that supports it.
Technologies like NX do a fantastic job of breaking the drawing down to cacheable sections. I am sitting on the slow-end of a remote link to my desktop at work as I write this and it is almost as good as being there. Latency is a bit of a bitch, but rendering and update speeds are awesome. I just wish that NX/X2GO supported GL better so that I could run the same desktop remotely as I do locally.
Microsoft does almost as good with RDP. I'm quite impressed with that even though it's a Microsoft abomination.
You do realize that the Wayland guys are also the core Xorg developers? They came up with Wayland because they realized X is not fixable.
X is a protocol for graphical interface elements, such as application windows. With remote X, the application's window IS on your local screen, using the remote cpu and fileystem. It's part of your local desktop, a real, local window.
VNC is a highly compressed PICTURE of a remote desktop.
Since X is the real thing, and VNC is a low quality PICTURE of what X is actually doing, it's just like you're saying that a porno mag is better than an actual girlfriend. Your comment is THAT ridiculous.
Besides the fact that you seemingly don't know the difference between an application and a desktop environment.
If you ever want to stop masturbating with VNC and try the real thing, use vnc -Y -C . Y is a better version of -X, and -C enables lossless compression, which is very useful on most networks.
It sounds like a line out of a Billy Joel song.
SSH with X11 forwarding isn't what many people consider top priority. The majority most likely go either full headless or full desktop. Everything in-between is likely to be an ad hoc solution to a problem that needs to be either headless or have proper remote desktop capabilities.
But nobody was talking about dropping the capability, that's plain stupid. Even though I use floppy disks only approximately once every two years and I don't regard it as a top priority feature, it doesn't mean I want the support for floppies dropped. However, it also doesn't mean it should be used e.g. as a data backup solution in a production environment.
X11 forwarding as a good-to-have option and as a doing-something-wrong method can exist at the same time. They're not mutually exclusive.
-SR
We do it ALL THE TIME where I work. We'll ssh from our Windows machines into the lab, do what we need, and if we need to open qtcreator or gedit or whatever, we'll run it and X-Win32 will serve it locally. VNC would mean importing a sluggish whole desktop which isn't required and given it's all bitmapped screen grabs will never be as fast as a locally rendered GUI, which is what X servers are designed to do.
FFS, so many people in this thread are totally ignorant about how X can be used and is used by people in the field. No wonder Linux is going to shit in certain areas - the lessons of the past are being forgotten.
There are a dozen use cases for not full headless and not full desktop. I'll name you one: a laboratory workstation that you both physically sit at and occasionally check up on from your desk or your home by sshing in and running a graphical thingie to monitor to test equipment it's plugged into.
And guess what: they did it wrong because they dropped a feature lots of people liked and use as part of their workflow.
And yet you can't be bothered to name at least one of them.
Il n'y a pas de Planet B.
So if you are on the same network as the machine you are connecting to, VNC is a reasonable option. With higher latency and lower bandwidth, it can get really painful.
The advantage of X is that you could only have a single application running over the network which tends to be much less consuming in network resource. Also that application would just integrate in your desktop environment perfectly.
Haven't tried wayland yet.
SSH with X11 forwarding isn't what many people consider top priority. The majority most likely go either full headless or full desktop. Everything in-between is likely to be an ad hoc solution to a problem that needs to be either headless or have proper remote desktop capabilities.
I don't know about the "likely majority"; but maybe you should research it.
I do know there are quite a few people who do use X11 forwarding. I do know that sometimes that just works so much better:
* alt-tab between different applications on different machines
* viewing two machines next to each other for reference (both data, and different architectures)
* cut-and-paste
VNC doesn't let the remote desktop interact with the local desktop. It displays it, but it doesn't interact with it.
Defenders have argued that network transparency is a minority application and that they don't like the way it's implemented in X11 anyway,
All I need to know is that all the people who know the most about X11 think Wayland is a good idea.
Here's a talk from 2013 where an experienced X11 developer explaining exactly what is wrong with X11 and why he thinks Wayland is a good idea. This link starts 40 minutes into the talk, where he specifically talks about running remotely over a network.
https://youtu.be/RIctzAQOe44?t=40m22s
And I've never seen a Wayland developer say that network transparency would never happen; they were focused on getting the essentials right.
Here's a talk from SCALE a year ago. This link starts with him saying exactly that: the Wayland guys were focused on essentials but now are ready to start looking at remote.
https://youtu.be/Sz1T0GvUziw?t=27m36s
lf(1): it's like ls(1) but sorts filenames by extension, tersely
Why not go whole horse and just use Windows? [...N]o remote terminal [...]
For extra laughs, Microsoft is actually working on that. Even Redmond has finally understood why some features are essential to an ecosystem even when only a handful of users need them.
Rudolf Hess edited Mein Kampf. He was the very first grammar nazi.
SSH with X11 forwarding isn't what many people consider top priority. The majority most likely go either full headless or full desktop. Everything in-between is likely to be an ad hoc solution to a problem that needs to be either headless or have proper remote desktop capabilities.
The majority of people don't know what's possible and what they are missing because they are pretty much self taught and never ever learned from others because they believe they already know everything. You don't need full remote desktops ever except for feeling familiar with them. For remote work it is much more work efficient to blend the remote desktop into your local one, because you often enough have to work at both (or more) places at the same time and full sized desktops just take too much space. The only time I really want to see a full remote desktop is if I can't figure out what the hell he is doing wrong at his station, but in this case I also want to see him push the mouse, etc. and then it's teamviewer time.
If you need to pull a full desktop instead of a single application window you aren't really flexible and your workflow has room for improvement.
When corporate requires Outlook, that full desktop is Windows. Then one uses remote X for everything else.
No, VNC is not an alternative, as it can not forward windows, only the entire desktop. Which interferes with things like switching between Windows (including Outlook), and getting the outlook notifications in the (currently visible) systray.
"ssh -X" works fine on wayland. Install Fedora 23, start a wayland session and try it. (Ok, it starts up an xwayland server underneath, but from a user point of view it works indistinguishably)
"ssh -X" already works fine on wayland. Install Fedora 23, start a wayland session and try it. (Ok, it starts up an xwayland server underneath, but from a user point of view it works indistinguishably)
It's not surprising. The people writing Wayland are the current stewards of X, and have been responsible for a number of really user-hostile things in X development too.
For example there used to be a keystroke for killing grabs. They removed it claiming it was "unnecessary" because you only need it if there's a bug in an application. So you should fix the buggy application, rather than just killing the grab (or application) and continuing with your day.
SJW n. One who posts facts.
Simple use case. I mostly program in Fortran these days, but I often use Matlab for smaller things. What i do then, is instead of installing the 6GB of Matlab and all its toolboxes on my poor laptop, I have it installed on my work computer. I then ssh in, use 'matlab -nodesktop' to start an interpreter, and use it interactively. Every time i use a plotting command, the plot window pops up nearly instantly on my laptop. I'm not interested in using VNC, because first of all it would be inefficient to forward the entire desktop when i only want to see a few plot windows, and secondly my laptop doesn't have the same high screen resolution as my work computer. I only want to see the plots, and 'ssh -X' gives me that.
What people want is ssh -X and yes it is a top priority to many.
That, plus the ability to reconnect to the same session (à la screen),
in case your connection goes lost for some reason, or if you want to
move to a different terminal (think remote/home work first via your mobile
device, then move on to your workstation as you get home after being
called when on call duty).
Similar as to what Sun did ages ago, with their Sun Rays,
of course updated and more flexible.
But that's not Wayland, it's X11, and only works for X11 applications. It'll stop working as soon as Linux applications transition to Wayland and the toolkits drop the X11 code.
>toolkits drop the X11 code I think that's a pretty long way off.
What utter bullshit.
I routinely use X forwarding on a 10 megabit LAN without any problems. More likely a poorly written application is to blame.
A really stripped down server only runs 1 tty, and forces the use of ed. Anything else is just fluff !
So if I want to install Oracle on a server, I'm supposed to fly over to a different country where the datacenter is located and sit at the console to do it?
He just realized to his shock that not *all* servers are webservers or DB servers and there are more stacks than LAMP.
Unicode killed the ASCII-art *
Who needs ed when you have cat ?
Unicode killed the ASCII-art *
Had they and their supporters not been so idiotic, Wayland would likely be the defacto Linux GUI today.
Errr no. Were Wayland actually ready for the prime time and not still a nightmare to get going it would likely be the default today. To be very clear many of us don't give a shit about network transparency but still can't get Wayland going for one reason or another.
I have always suspected that the slowness of X11 over a network was primarily latency, not bandwidth. I think it waits for acknowledgement after each step before moving in, rather than (ironically) windowing the traffic. Paradoxically, adding compression to the mix may actually make things worse, because the compression and decompression steps, even if they keep up with the bandwidth demands, will add to the length of time it takes to get a request fully received, acted on, and acknowledged.
It's just a suspicion, though. I could be completely wrong.
www.wavefront-av.com
Yeah, that should be ssh -YC or ssh -Y -C, not vnc.
(By habit I normally seperate my flags since -AB can mean -A B, with B being an argument to A).
KVM switch is cheaper than flying. Any respectable datacenter has them.
-SR
Is Fedora 23 a "modern Linux distro"? If so, to use Wayland on a modern distro, click on the little gear under your name on the login screen and choose "Gnome on Wayland". It's so easy I've done it by accident. (Synergy still has no Wayland support, so I don't want it as my default, but GDM remembers what desktop you chose last time.) This is a "real pain in the ass"?
My current complaints are that Synergy doesn't work, which isn't really a Wayland failing at all; lack of an xrandr equivalent that I've found in a couple days of casual looking; and that I can't middle-click to paste anymore in native Wayland applications. (That last actually does still work between applications running on XWayland.) The availablility of XWayland also should mean that you can still use ssh -X from a desktop running Wayland and forward remote X applications. Still, I'm happy to see networks transparency developed for Wayland, because eventually X won't cover everything graphical I want to run from a remote VM.
Wayland already handles multiple monitors with different DPI much better than X, which is why I went ahead and tried it for a while recently. Other than the items mentioned above, the experience was barely noticeable. If I can find a replacement for xrandr to use in my screen rotation script, I'll probably switch my convertible laptop from X to Wayland by default, since I don't use it with Synergy and I do plug it in to a variety of external displays. My work laptop and desktop systems will probably stay on X until there is Wayland support in Synergy, or a Synergy replacement for Wayland. For my use, I don't even care if such a replacement supports any non-Wayland displays, since I could switch everything at once and non-Linux OSes are confined to VMs in my life.
Citrix, VMWare View, etc.
Support my political activism on Patreon.
Which works fine if your equipment supports multiple sharing sessions. If not, starting the new monitor may disrupt the existing process, screwing you over. Which is why X and remote desktop are NOT mutually exclusive - sometimes a view-only session is all you need to quickly view a setting without running something that could disrupt your long-running process.
The other reason is if you're on a flaky connection. Do this and X becomes a poor solution because the moment your connection burps, your applications are force-quit and you lose your work.
There are situations where one solution is better than the other. X forwarding is great, but it's not the be-all-end-all solution, especially if something you're doing is single-session only or you're not on a reliable connection (e.g., mobile) where you don't want all your programs to abort because you lost your cell signal.
(By habit I normally seperate my flags since -AB can mean -A B, with B being an argument to A).
With separate flags that becomes "-A -B", with -B being an argument to -A. How is that better?
"The state is that great fiction by which everyone tries to live at the expense of everyone else." - Bastiat
For example there used to be a keystroke for killing grabs. They removed it claiming it was "unnecessary" because you only need it if there's a bug in an application.
They removed it because it was a security problem, not because it was "unnecessary". You could use it to bypass lock screens, which are implemented in part through screen grabs.
The AllowDeactivateGrabs and AllowClosedownGrabs options are available in xorg.conf if you want to restore the original insecure behavior.
"The state is that great fiction by which everyone tries to live at the expense of everyone else." - Bastiat
What people want is ssh -X and yes it is a top priority to many.
That, plus the ability to reconnect to the same session (Ã la screen), ...
In other words, what people really want is the functionality provided by xpra. The thing is, xpra would actually be easier to implement as a Wayland compositor than the current hack based on Xdummy or Xvfb.
"The state is that great fiction by which everyone tries to live at the expense of everyone else." - Bastiat
I routinely use X forwarding on a 10 megabit LAN without any problems. More likely a poorly written application is to blame.
The problem is that an X application which is written correctly for local display (for example, taking advantage of hardware acceleration) is "poorly written" for running with a non-local X server, and vice-versa. To handle both cases well you have to implement two different UIs, which shows that X's much-vaunted "network transparency" isn't actually transparent at all.
"The state is that great fiction by which everyone tries to live at the expense of everyone else." - Bastiat
I think you accidentally ticked that "Post Anonymously" box.
-SR
Most Linux programs use the getopt() function from the C library.
getopt() in glibc treats -A -B as two switches. -B can never be interpreted as an argument to -A. On the other hand, -AB DOES set B as the argument to -A, if -A can take an argument. -AB is two arguments IF -A can't take an argument. So on Linux, -AB can have two different meanings. -A -B has only one meaning, it's always two switches.
http://linux.die.net/man/3/get...
An example might be helpful. These two commands mean completely different things:
perl -i -w
perl -iw
The first means both -i (inplace edit, do not create backup files) and -w (show warnings)
The second means inplace edit, with backup files named "w". (Do NOT show warnings. )
If you mean to pass two flags, -x -y will always do that. -xy sometimes will, but sometimes it has a completely different meaning.
Right, because they haven't learned anything whatsoever in 30+ years of developing and maintaining a massive codebase of extremely widely used software.
Apparently not: they're blaming X (the protocol) not Xorg, their implementation of it. Bear in mind under their stewardship it's been largely rewritten. Some of the mistakes are theirs and they're blaming others.
That does not sound like a bunch of people who have learned lessons and will do it better next time.
Compare that to the slow and carful translation of GCC to C++ along with refactoring.
SJW n. One who posts facts.
Avoiding inconsistent behavior due to optional arguments is the whole point. Reread what I just said:
-AB DOES set B as the argument to -A, if -A can take an argument. -AB is two arguments IF -A can't take an argument. So on Linux, -AB can have two different meanings.
It's easy to forget that -A CAN optionally take an argument, if you're accustomed to using it without. You wouldn't forget that -o always has a required argument, if you use it at all. That is, you'd never use "ls -I ht" expecting it to behave the same as "ls ht -I", because obviously -I needs argument.
You're moving the goalposts. What you said was:
So on Linux, -AB can have two different meanings. -A -B has only one meaning, it's always two switches.
"-A -B" is two switches only if "-A" does not have a required argument, otherwise it's one switch. It is not true that "-A -B" is always two switches.
If you're not sure whether a switch takes an optional argument, then the "-AB" and "-A -B" forms have the minor advantage of being unambiguous given that the switch either can take an argument or can be used without one, respectively. However, a better solution would be to consult the --help text or manual page and remove the uncertainty.
"The state is that great fiction by which everyone tries to live at the expense of everyone else." - Bastiat
> However, a better solution would be to consult the --help text or manual page and remove the uncertainty.
While carefully reading the man page again for each command you type and carefully comparing what you're typing for each option to it's manual page specification -would- be effective, I'm not sure how many people want to spend the time doing that. Tapping the space bar by habit is a MUCH quicker way to avoid the common error case.
Much like consistently putting switches before arguments:
command file1 file 2 -a -b -- Sometimes works as expected, common source of error.
command -a -b file1 file 2 -- Consistently works as expected, rarely / never causes error.
It doesn't take any longer to type the options before the arguments, and it works with POSIX-correct software.
I use it to access LibreOffice instances running on remote AWS servers. It's perfectly fine.
It would be nice if X11's critics actually used X11 once in a while, rather than whined incessantly about how bad it is because in theory it's bad.
You are not alone. This is not normal. None of this is normal.
Aside from all the jokes about not using vi, etc. I seriously marvel at the idea of a "really stripped-down server" including an X system. Though I fairly regularly use ssh -X on other workstations, I haven't included X at all on a single server (LTSP servers aside) in the last 15 years. I'm curious what kinds of tools server admins are requiring an X environment on the server for.
Actually, after writing that, I realized that, yes, really a stripped-down server does not include X. I have some servers configured this way.
The point that I was trying to make is that a server should not need a desktop window manager. Earlier, in the arguments over Wayland, the response was "just use VNC". But for VNC, you need a window manager.
I have found that managing KVM guests without virt-manager (and hence X) is a pain (although not impossible), so on my fairly stripped-down servers that act as KVM hosts, I include virt-manager and X. I run virt-manager as a network-transparent application, with the X-server on my desktop. On the servers that don't host VM guests, I can eliminate X.
The real "Libtards" are the Libertarians!
That's a good point -- while rare for me personally, I have on occasion included minimal X libraries for particular apps before while not hosting a full-time console environment. I've run into some apps that offer GUI admin from a remote machine -- but only if you install a "big boy" web server w/ extra modules that starts to make the X installation look small and simple.
Wayland can do everything X does but X cannot do everything that Wayland can. Network transparency is a low priority feature compared to getting core features and cross-platform implementations working and stable. If you want it, add it yourself. Isn't that the opensource motto?