Voice of the Masses: X/Mir/Wayland – do you care?
Before we get to this episode’s big topic: we’re delaying the podcast by a week, so that we can all get together in England and meet up with the generous buyer of the £1,337 “Meet the team” meal perk from our Indiegogo campaign. It’ll be great to meet him, and chow down on some lovely grub in Bath!
Anyway, back to the matter at hand: there’s been a lot of discussion about display servers recently. As you’ve probably seen, X is on its way out (at least on typical desktop PCs), likely to be replaced by Wayland and/or Mir in time. Some developers argue that the display server doesn’t matter, because application developers are using toolkits that add a level of abstraction, whereas others say it does matter because the underlying behaviour changes. (Also see Aaron Seigo’s take here.)
So, we want to hear your opinions. Do display servers matter? Do you care about the technical details of the implementation? Or do you think all this arguing and duplication is fragmenting the Linux and FOSS Unix desktops, and wasting too much time? Let us know in the comments…
Related Posts
About The Author
Mike Saunders
Mike stores his data by printing out hex dumps and laminating the sheets. He's the author of The Haynes Linux Manual and the creator of MikeOS, an operating system hand-built from assembly language.
Although I listened to a lot of different linux related podcasts talking about the subject of display servers, I still don’t know the exact advantages of Wayland and Mir over X (and even more the differences between Wayland and Mir).
What I think is that Canonical has achieved a lot in making Linux more usable for the masses. Canonical made it usable for me as well. I tried using Linux (Suse) in the late 90ies and was not able to make it work for me as my main everyday OS. That changed in 2008 when I tried out Ubuntu 8.10 Intrepid Ibex. I was very happy and surprised how good it could do anything except running some games that my kids like to play.
So coming back to the original question: I don’t care too much about which display server Linux distros are using, as long as the maintainers work hand in hand and Canonical doesn’t pack in and leaves the community alltogether.
I am no expert by a long shot, but apparently Wayland could prove more resource efficient. X is also a mess. It is a working mess that we love, but still a mess.
Mir seems to be a way for Canonical to control the commercial phone and desktop “market” which is a hybrid approach ; proprietary until after Canonical has made money on it. That is a political and moral problem rather than technological.
The “replace X” to modernize the architecture critique is well founded. Software paradigms have shifted, for better or worse. Mir is simply business strategy, however.
i hope there is no agreement on one single way of doing a task. think of it as having a bet on 3-4 horses in a race. if one falls for some reason you’re still looking good. the competition and innovation shared between them benefits us better than having one single stagnant method.
That depends on how the different horses interact. That more alternatives is better from the outset, is untrue.
There is no magical positive element of more choices if all those choices are wrong. It depends on opposition and interaction.
Be careful what you wish for. X11 embodies years of requirements. Yes, it also includes a bit of “baked in” cruft. Wayland, as a poster child for re-factoring some of the cruft out of X is a fine ides, but he “sales pitch” for wayland is absolutely dubious. You’re focus on “choice” works out like this…
1) When you write an an application, any application, you aren’t targeting a “standard”, like X, you are targeting a wayland compositor. Choice is good, eh? So, if I may ask, which of the dozens of compsitors will, say, Firefox code to? How about LibreOffice? Chrome? Oh, you want to use all three? Too bad for you. You get your wish, choose one, any one.
2) Wayland devs say Window Decoration is a client side issue. Good lord. Load 5 Applications, each with entirely different title bars, entirely different resize/menu/functional semantics. Fabulous. Just freaking fabulous.
So… “There be standards for that!” Yea, no. “Standards” have proven themselves not worth the paper their printed on. “Glibc can’t even keep a stable interface” (Torvolds). The only standards that have any hope of sticking are those encoded in software and defended by their authors. Like,say. those X11 forces upon programmers.
Wayland is great for embedded devices, closed ecosystems. Only there can any sort of global standard be enforced, if for no other reason than the threat of getting fired.
X11 could use a re-factoring. That it’s “bloated” is a lie. Right now Firefox, non-mixer, and mplayer on my machine are all consuming far larger resident set sizes than Xi is. That its “outdated” is also a lie, just as it would be to call Toyota the same.
The only legitimate complaint I’ve heard about X is the need to support whole useless functionality like some of the ancient rendering plugins. Ok, so pull them. Fix things that break, lather, rinse, repeat.
I think that this is a really important issue for people developing things like graphics libraries etc. However, for ordinary developers and end users I don’t believe it is something we should get too riled up about. If it was going to change the UI that faces me every day or drastically alter application performance then yes, lets have a debate about it, but for now I think we should be happy to let Canonical or whoever do their stuff and wait for our new favourite Linux magazine publish a “Mir vs Wayland vs X” article in a few months time.
I totally agree! This issue has little to do with end users. We will end up with feature parity or surplus, thanks to X providing a fallback.
When I hear sysadmins boasting their experience and then swearing to X, it is apparent that they are talking about whatthey know. What they don’t know is that X is considered sspaghetti code by those who wrote it.
Wayland matters, and it’s development has brought many improvements to all areas of the graphics subsystem, even those used by X and Mir.
Mir seems pointless, it was developed because Canonical misunderstood how Wayland would work, and they managed to annoy the Wayland developers when they released it because of their inaccurate accusations.
X is obsolete and broken, it isn’t even truly network transparent any more. Here’s a good video by an X developer as to why we need Wayland: https://www.youtube.com/watch?v=cQoQE_HDG8g
It doesn’t matter what the name of the server is, but it matters that many had already began preparing for Wayland when Canonical started their Mir project. Wayland is already in use in some devices, for instance the Jolla smart phone. This is a device that you can buy today! Let Canonical try, but they won’t have the support and trust of the free software community (not as much as Wayland), because of Canonical’s policies. But they are free to try. Diversity is good up to a point. But we saw what happened with the Systemd vs Upstart debate.
Contrary to popular belief,Canonical is also part of the free software community. There were a lot of benefits to upstart that are lost with the migration to systemd. Mir borrows a lot from Wayland, and there is the potential for Wayland to borrow from Mir too. The trouble is that there is a strong anti-Ubuntu feeling fostered by flame wars, despite it being the one distro that can be said to have truly brought more people to Linux than any other.
That anti-Ubuntu sentiment was fostered by Canonical themselves, because of they way they announced Mir. If they had handled it as part of the free software community instead of behind doors while everyone was waiting for the promised committment to Wayland, we’d have little to complain about.
Instead they did all that, and then to add insult to injury came up with completely inaccurate “technical” reasons for not using Wayland that were quickly stripped down by people who knew what they were talking about.
They pissed a lot of people off, perhaps not your every day users, but more involved people in the free software community.
(I hope this reads well, I’m too tired to proof read)
This the point, Rowan. Personality clashes and words just divert attention from the actual facts. If people react emotionally rather than objectively, then you miss the reason why one group went one way or the other. The open source community’s biggest problems are the passionate personalities involved who are quicker to react than reflect…
I agree, it was highly emotional,I was just attempting to summarise how the negativity came to be. You see a lot of this any time Lennart Poettering is mentioned.
However, I’m not convinced that there’s any objective reason to consider Mir over Wayland, but both of them are for the most uses objectively better than X.
Hi
Personally I dont really care as to what display technology various distros use under the hood so long as it works. My main bugbear over the years was trying to configure Xorg to display Mythtv on my TV using TVOut, thankfully this is all a memory now.
I would prefer though if all the distros would standardise on one technology though 🙂
As a user no, it does not matter.
Just read Wayland and/or Mir as KDE / Gnome.
Does it matter?
Mir vs Wayland is just another Canonical/Red Hat bitchfest gone viral but it is so last year… I can’t believe you people are even talking about it. Surface Flinger won this battle for an x.org replacement way back in 2009…
Latest bitchfest that everyone’s onto is Mark Shuttleworth vs Jon Masters on ARM64 ACPI vs Flattened Device Tree. Get with the program Linux Voice and try to stay current.
I’m an elderly user. Simplicity and out of the box functionality matters….Wayland vs Mir doesn’t to me, as long as it works.
I agree with you sir. as 77+ out of the box matters more than ever now. I used to program in basic but I’m afraid all that is lost in the mist of time now. Thanks for bringing this up. Vic
Yes I do care as I use things like the ssh tunnelling of X a lot when using my Pi’s or headless server.
Having looked into it alittle as a point of interest, I found I agreed with the points raised by Wayland. The way X handles modern feature is clunky and slower and did need re-visiting.
Mir seems to be Canonical behaving like Sony did with blueray and not wanting to co-operate with Wayland to produce a single all singing all dancing solution, but instead muddy and confuse the issue.
This has, in my opinion, been a developers argument more than a users one.
Noone at Canonical or Ubuntu has, as far as I know, ever said Wayland is not the way to go for everyone, they just said it wasn’t the way they wanted to go.
As long remote graphical sessions work, as long as programs on the desktop and server continue to function and as long as apps on the tablets and phones perform correctly, who cares at a user level?
It matters to SOMEONE, but that someone is not me. In general, its not going to matter as long as it works.
There was a good talk at Linux Fest Northwest 2013 broken into 2 sessions, ‘Why Linux Sucks’ and ‘Why Linux Doesn’t Suck’. The presenters did a great job walking though all the points (things like decentralization and fragmentation) from both perspectives and showing that no single opinion matters in the least. Its the fact that the smart people out there are making great code for other smart people to integrate and make it ‘just work’ for the masses.
So ‘Thank you’ to the smart guys that are still working the X project, as well as those working Mir and Wayland. And ‘Thank you’ to the distro developers that work with them to make it ‘just work’ for everyone else who just doesn’t have the time for LFS.
This is a technical, behind-the-scenes, issue which seems provocative more to developers than end users. As an end user myself, I can see how old technologies need replacing as systems and software evolve. If Wayland is the emerging new industry standard for graphics servers, that’s fine.
If Canonical wants to go their own way with Mir, then isn’t that their choice?
Presumably, if Mir were to fail for some reason, that would mean wasted resources and loss of credibility for Canonical. But what is the consequence of success? Are people going to choose a Canonical tablet/phone/TV/desktop on the basis that it has Mir instead of Wayland?
So I don’t see how it matters to the industry in general. I must be missing something.
For a user like me, the distinction is theoretica. If my distribution offers both, I will try both. It seems fair that X was started in a different time, and it would be time to consider next generation instead of next iteration, given a very different environment.
I rather like X for its cross-platform, and remote accessibility features. So long as it’s successors do all the same things, but better, I’ll be happy.
Of course it matters, of course we should care. But not for the obvious reasons. Divergent mutation is necessary for evolution. When going from one tried, tested, stable platform to a revolutionary one, a period of experimentation, trying different and if necessary divergent approaches, before borrowing the best of both approaches, converging and settling on the ideal replacement is key. The final platform is immaterial. The fact that there is this natural selection process is vital.
This is open source we are taking about, folks. Open source non-commercial, competition is cooperative, by its very nature. This is one sport where you can support both sides and whoever wins, every one wins.
Waylandist developed by intel and pushed by redhat, so wayland imho is being influenced by big money, To me it looks like intel and red hat are finally trying to take over linux, which is a very very bad thing imho, especially with intel trying to kill off dektop computing and red hat trying to create GnomeOS as de-facto linux standard system. What will be left when intel has killed off desktop computing? Gnome Waylinus OSfor mobile devices only, so desktop users might be forced to go back to Windows Wayland sucks big time imho, I refuse to use it and I will turn my back on linux after a decade when wayland destroys Linus as we know it.
Die Intel, die red hat!
The X and Wayland comparisons and benefits are still a current topic of interest. As for staying current and following trends. The Linux and Free Open Software movements are about having options, not just following trends.
My above comment was regarding
spych102 !
As a general desktop bod, as long as whatever is replacing X is ready to replace X then I shouldn’t even notice a difference (performance improvements aside).
X is at times a bit cumbersome but does work on the whole. Whatever takes over needs to be better than X at release, not launched on the basis that it has the potential to be better than X.
I for one has not seen any distro, using surfaceflinger, might there be a reason?
Even thoug I am a GPL fanboy in some cases I can see benefit for a less restrictive licence. In this case (MIT) is reason enough to choose wayland as the new defacto displayserver. Users of all operating systems in existence would be able to partisipate and develop it.This would have been a continuation of the way X11 have been used and developed over the years. If say all linuxes ralleyed behind Mir it would probobly not receive contributions from outside the Linux sphere. This problem beeing exagerated by Canonicals contribution agreement.
I’ve read a couple of articles about Wayland/X/Mir and tried to follow the discussion on a arch.reddit.com whenever those articles are posted, but as a non-developer most of it dosen’t make a lot of sense. I seems like X has a long history, and thus a lot of legacy support, and Wayland/Mir is the preverbial clean slate. I will try Mir/Wayland out, if it has better preformance, and properly will anyway just out of curiosity. More importantly is properly the we are given a real choice about our display server. The software will get better with “compatition”, or at least there is a bigger change “special feature X” will get implemented in one or the other, and thus give users more choice, which we know we all love.
I personally can’t wait for Wayland to replace X. X is the wrong architecture for handling the display in this day and age (you could also question if it was the right architecture during the 80’s). Each extension to X has moved further away from the original approach (eroding network transparency as it went) turning X into a pile of extensions all hamstrung by a big pile of messy IPC. Xorg has done a very good job of clearing out alot of old cruft and just pointless junk that has no place in the display system (does any body remember X print), however you can only go so far with this approach. The architectural issue with X still go unsolved thus Wayland.
The performance issues in X causes al-sorts of problems for the modern Linux desktop that no tool-kit or desktop can completely hide or workaround so replacing this layer will make all Linux desktops run smoother and faster.
One of the main concerns for some users in the move away from X is the lost on network transparency, however plent of X apps are not that network transparent and RDP will bring proper network transparency to Wayland.
I’ve mentioned Wayland quite alot but not Mir as Wayland has the momentum and cross distro support to become the de-facto replacement. Mir on the other hand is a good solution for Canonical’s needs to create a mobile/desktop product but at it’s current rate of the development and distro adoption its unlikely to ever spread outside of products directly controlled by Canonical.
Latest News Flash folks!!!…the debate is OVER…Canonical and Intel have announced that Mir is to be renamed Wayland+. This will create a Wayland compatible display manager and merge the efforts of the two projects with a truly platform independent solution! Head over to Omgbuntu for the announcement by Shuttleworth!
@Matthew S.
But Matthew Android uses Surface Flinger and Android is almost all of Linux screens. Mir and Wayland are not terribly relevant until they prove themselves technologically and on top of that take some actual market share.
As a user I care about two things:
1. Hardware compatibility. The single most worrying thing to me about the switch away from X is that it has the potential to remove or reduce compatibility with older hardware. The ability to repurpose older hardware on Linux is important to me and many others.
2. Software backwards compatibility. An X server layer is not neccessarily sufficient as it only addresses the “application” side; what if I want to continue using my old Window Manager, for example (there are some other examples too)? In this case I’d need to rely on someone porting it which may not be an easy task and may not happen for older/less popular software.
These are both issues no matter which of Wayland or Mir “wins” so I am agnostic between the two.
I’d add that to me, network transparency is a red herring, as long as there is a solution which works for network display – over the internet you currently need to resort to things like FreeNX to make X usable, so if anything this is an opportunity to do network display much better than X does by making the protocol more high-level (e.g. like Windows RDP does it).
As with other commentators I can’t say I’m particularly bothered either way. As long as whichever system is in use doesn’t require some stonking GPU I don’t care. I just want my desktop ( XFCE/Gnome3/KDE/enlightenment/whatever-I-fancy) to work smoothly on whatever distro I choose and whatever hardware I have to hand, be it a Dell laptop or an HP Proliant server (not that I run graphical interfaces on servers much!
I frequently work from home and use X redirection for some applications and Nomachine to gain access to my desktop at work. Providing both these methods work, I’m happy.
Anyway, isn’t this XFree86 vx X.Org all over again? 😉
I suppose developers may have a preference but, as long as it works well and efficiently, who else will care.
The problem as I would see it would be if both systems become established as this would fragment the work of programmers. Perhaps there will be a need for a tool kit to allow Wayland to work in Mir and visa-versa.
This just in! There is still no answer.
I matters to me as I want an alternative to blacklisting nuveau drivers just to get my Nvidia card to work.
I don’t want to have to dive into the internals of the OS just to get 3d from my video card.
Sure there are some very good tutorials, but like I said, that is not what I need…I just want it to run out of the box and to be able to choose “non free” drivers.
So far NONE of the distros that let you choose non free will work with my so called legacy hardware (Nvidia 8600gt) which is not that old.
Since systemd will force its future graphical stack on everyone, any and all other alternatives are irrelevant. 😀