KWinFT Lands Code To Now Use WLROOTS For Wayland

[kpedersen]

That has not gone anyone where in 6 years. There is a reason.

I am sure there are many reasons. One could simply be that Xorg is currently going quite strong so Xweston isn't really needed. However this could change as Wayland becomes a more common default.

As for the technical reason you mentioned, I certainly won't deny that there isn't a disconnect between the two but I don't see why this can't be worked around. Or will the majority of guys just ditch their window manager they have been using for decades and just make do with Gnome, KDE or Sway? I am not entirely convinced.

[oiaohm]

As for the technical reason you mentioned, I certainly won't deny that there isn't a disconnect between the two but I don't see why this can't be worked around. Or will the majority of guys just ditch their window manager they have been using for decades and just make do with Gnome, KDE or Sway? I am not entirely convinced.

History of the changes between KDE and Gnome being dominate and back again that happened many times. Majority of Linux users are not bound to their windows manager. This comes clear I have KDE on computer, XFCE on another in the same house and I use both computers. So I finally get around and change XFCE to KDE I am not going to notice any usage difference right. The majority from the historic moves tell us that the majority of Linux users don't care that much about their desktop environment other than it works well. This is around 90% of the Linux user base and that is from the historic moves from KDE to Gnome making up like 99% of the desktop then to like what we have today. Yes this pattern historically has cycled many times over the past 20 years. Really I don't see why this time will be any different. Yes one of the historic ones was caused by a X11 change in clipboard handling resulting in Gnome and KDE working well and the other windows managers not and in 2 years KDE and Gnome was like 99% of the Linux users as the other caught up the diversity returned.

I don't see why this can't be worked around.

The problem here is mutter developers in early Wayland support did attempt to work around it. Why this was so they would not have remake as much code. So we have a functional example the nightmares here.

[t.s.]

The majority from the historic moves tell us that the majority of Linux users don't care that much about their desktop environment other than it works well.

And have many advanced features. I've tried using XFCE and E24. But back to Plasma because it have more features, better integrated with services, and their notification system's better. Oh, and the theme too. KDE theme configurability's superb.

[tomas]

bug77

I'm not sure wlroots has proven itself outside of sway

Yes it has. It is also used by Phosh:

https://honk.sigxcpu.org/con/phosh_overview.html

Here is a complete listing of projects using wlroots:

https://github.com/swaywm/wlroots/wi...ch-use-wlroots

None of them can be considered to be major ones though. I think sway and Phosh are the most active. ones.

[bug77]

bug77



Yes it has. It is also used by Phosh:

https://honk.sigxcpu.org/con/phosh_overview.html

Here is a complete listing of projects using wlroots:

https://github.com/swaywm/wlroots/wi...ch-use-wlroots

None of them can be considered to be major ones though. I think sway and Phosh are the most active. ones.

I disagree. I've heard of only a handful of projects on that page. And the page includes KwinFT, which has only started to use wlroots. That to me is not proof of maturity.

You see, what I'm looking for is validation on major projects which usually guarantees you're free of idiosyncrasies from the first thing you supported. Think about how Rust has exposed C-related idiosyncrasies in LLVM. Bear in mind this is in no way a criticism towards wlroots (or LLVM). It's just how I've learned things happen after years in the industry.

[tomas]

bug77
I'm a bit surprised you have not heard of Phosh. It is used both on the Pinephone and on the Librem 5. But I agree we're not talking about a major project compared to KDE or Gnome.
Well, only time will tell how well wlroots will mature. My view is that it's already showing its strength outside of Sway.

[oiaohm]

I disagree. I've heard of only a handful of projects on that page. And the page includes KwinFT, which has only started to use wlroots. That to me is not proof of maturity.

To be correct most people have only head of a handful of X11 windows managers. And there is not as many of those as what you would think 41
by Wikipedia and 110 by xwinman site Yes if you add on DE that is only another 20 from xwinman bring that total to 130.

https://www.gilesorr.com/wm/table.html

This site here attempted to have a total list of all X11 windows managers that have ever been existence this taps out at 499 including dead ones please note this count here includes display environments as well.

This puts wlroots 43 know users in a complete different light. This is coming up on 1/10 the number solutions that X11 has got over its complete life time. What X11 tells us is there is roughly at 10 to 1 if you don't go and look up the list know existing/have existed windows managers. Meaning you roughly know about 1 X11 WM/DE for every 10 that has existed. So only heard of a handful exactly aligns with this general rule since 43 means you should only really know between 4-5. Please note 43 is well ahead in the count compared to libweston and qt for doing wayland compositor heck this is ahead of write you own Wayland solution(yes that is ahead of libweston or qt usage).

Yes the fact the ratio is right with wlroots list suggest is a true and correct list to a point. Please note I said a point if you look at that list in the wayback machine you will see removed wlroots using compositors that for the gilesorr style list would have still been left in the list.

https://wiki.gentoo.org/wiki/Wayland_Desktop_Landscape

This is a interest read.
3 out of the 6 marked as daily drivers are wlroots. Yes Hikari, Sway and Wayfire are all wlroots. The two in beta are Cage and Cage Break are wlroots. The expermental DWL, Kiwmi, labwc, River and waybox are all wlroots. So that 10 out of a list of 15.

KDE and Liri are the two Qt based ones on that list.
weston of course is the libweston single showing. (we know there is more libweston based solutions in "Automotive Grade Linux" and so on but we are not seeing libweston base appear in distributions targeted at general desktop usage)
Taiwins started with libweston now doing its own thing.
Enlightenment of course is doing it own thing.
Mutter being gnome has been doing its own thing.

This is gentoo but the same trend is across most Linux distributions implementing wayland that the majority of the alterative options being offered by distributions is end up being wlroots based.

You see, what I'm looking for is validation on major projects which usually guarantees you're free of idiosyncrasies from the first thing you supported. Think about how Rust has exposed C-related idiosyncrasies in LLVM. Bear in mind this is in no way a criticism towards wlroots (or LLVM). It's just how I've learned things happen after years in the industry.

Sorry validation by major projects has not seen X11 protocol get free of idiosyncrasies as anyone who has done support with wine with people deciding to use non common windows managers know..

The reality by what distributions are accepting wlroots is showing usage outside sway.

Something to remember here the major projects not using wlroots are mostly doing their own thing. They were mostly doing their own thing in X11 time frame driving projects like wine nuts with cases where gnome, kde and enlightenment had decided to implement things differently like drag and drop.

Yes there has been a hold up with RHEL and Debian accepting lots of wlroots based due to wlroots/sway refusal to have anything todo with Nvidia Eglstreams but that is about to change that why I said most distributions.

Also do note X11 windows manager are broken into 4 common types of windows management tiling, stacking, dynamic and Kiosk wlroots has functional examples of all 4 being implemented. Gnome with Mutter, weston, Enlightenment backends in fact don't have examples of all 4 types. Yes Taiwins went it own way because libweston in it current form is not suitable for dynamic . Taiwins we are not seen other Wayland compositors bases off it.

Qt does have examples of all 4 types of windows management being implemented as Wayland compositor. So far there are only 2 all rounder toolkits for making a Wayland compositor that been qt and wlroots. Yes wlroots is winning the market share in the alternative implementations. Wlroots has about 90 percent market share in the minority options for wayland compositors this is why when you see someone attempting to make a list of wayland compositors you end up with majority being wlroots based over and over again.

Yes the fact wlroots has been able to implemented as the 4 common types of windows management is shows it is limited on idiosyncrasies just doing that.

Bug77 this is the thing major x11/Wayland desktop environment projects have more resources so they can do their own thing this has been the case since basically forever. The minor projects are the ones that have had to pool resources or die. Its really simple to miss how large wlroots market share with developers of alternative options has got because its not one of the majors doing it. Something else to remember with X11 history there was a time frame when Gnome, KDE and Enlightenment combined had reduced to less than 10 percent of Linux desktop users with the 90 percent being broke up between 30 alternative options nothing says this cannot happen again. Yes that was caused because Gnome and KDE and Enlightenment all did a screw up in implementation at the same time. Wlroots is currently well placed to take advantage of such event happening again.

[Morty]

Please note 43 is well ahead in the count compared to libweston and qt for doing wayland compositor heck this is ahead of write you own Wayland solution(yes that is ahead of libweston or qt usage).

Or if you look at the number of users and age of the compositors, those 43 is microscopic. For instance Lipstick, Qt based and been used for years with paying customers. Perhaps you have not hear of it:-) And then you have all those WebOS LG TVs, shipping with their Qt wayland compositor for years.

(we know there is more libweston based solutions in "Automotive Grade Linux"

And we also know most automotive is done with Qt, so I'll guess there are some Qt bases solutions there too.

[oiaohm]

Or if you look at the number of users and age of the compositors, those 43 is microscopic.

This does depend what you are looking for.

For instance Lipstick, Qt based and been used for years with paying customers. Perhaps you have not hear of it:-) And then you have all those WebOS LG TVs, shipping with their Qt wayland compositor for years.

These are not general desktop compositors. Majority of wlroots ones are in fact targeting to be general desktop compositors.

And we also know most automotive is done with Qt, so I'll guess there are some Qt bases solutions there too.

This is not as great as one would presume something really odd has happened. Automotive Grade Linux is one of the majors it compositor is libweston based but applications are majority qt based. Turns out when you get into cars and the like more often than not the compositor is libweston based even if the applications are Qt based. There has been major problem validating the qt implementation of wayland compositor to the standards automotive needs. Its very interesting that compositor here has gone libweston with the applications being qt I will give you Morty its not what people would expect with how heavily qt applications are used in automotive.

libweston has a decent market share in you need a validated low feature Wayland compositor market at over 90% of it. These compositors have a decent amount of age on them. So by number of users if you count all car users libweston beats everything. But these are not you general desktop usage cases.

libweston, wlroots and qt when it comes to wayland compositors if you are counting by number of users libweston, qt then wlroots. If you are look at number users restricted to just the desktop use case it qt(mostly just kde), wlroots then libweston. Where is our alternative desktops likely to come from wlroots.

I should have put Linux desktop distributions instead of just Linux distributions in quite a few places.

Yes minority who use alternative desktop implementations on desktop Linux distributions are quite microscopic over all. But this has been source of diversity to the Linux desktop distributions. At this state the lead item to keep them alive is wlroots. Number of users and age is only part of it the other is use case.