To be clear here, we don't have a compositor yet. I've done a little work on it, but it won't be usable for at least a year, maybe (probably) longer.
4.20 will let you run some of our components on another compositor. We mostly test with Wayfire and Labwc, as our compositor will also be using wlroots.
Awesome! A lot of people are excited, XFCE is a very charming and beloved desktop environment. I'm curious to see how people will take this change though... I know a lot of XFCE users either want Wayland or don't care, and then there are a lot who will probably be very resistant to this change. Taking it slow at least allows you to take advantage of the recent growth of Wayland and allows you all to polish XFCE Wayland up so that there are minimal hiccups during the switch. I'd be curious to know what your opinion is on the matter.
It's not so much a change as it is (well, will be) an option. We're not abandoning X11, and I expect it will be a couple more major releases before we have feature parity on Wayland with what we have in X11.
But yes, some people are excited, some don't care, some are worried.
But there won't be a switch. If users want to keep running it on X11, they can do that. I can't promise that will be the case forever, but it's not (won't be) that hard to support both with how things are currently.
(uninformed outsider) have you considered reaching out to the System76 folks? They are also setting out doing a Wayland compositor for a (decreasingly) gtk-based desktop and are reasonably far along on this goal. I wonder if their compositor would suit your needs?
To me, at least, xfwm4's window management behavior is a big part of what makes Xfce... well, Xfce. So no, I don't think I'd consider using someone else's compositor.
To add a bit, hours later: I've been testing my Wayland support for the other components using Wayfire and Labwc, and honestly I find their window management incredibly frustrating. Part of it is just lack of familiarity, I'm sure, but honestly I just don't like how they do things.
A lot of xfwm4's behavior is windowing-independent, or fairly easy to put the windowing-specific calls behind interfaces. There's some of it that turns out to be pretty hard to untangle, which is where the real work is, but I think it's worthwhile work to take on.
Sure, the xfwm4 behaviour is absolutely something you'd need to be able to own in a technical and organisation sense. However all the other stuff around building a Wayland compositor is... big. Sharing portals, XWayland support, high dpi and mixed low/high dpi, display management as a whole, and that's just the stuff I know about..
I love xfce, I don't use it anymore though because I wanted Wayland support. As a (past and potentially future) user, I want to see a healthy xfce, not one that is burning out its devs with the giant task of maintaining a Wayland compositor. :)
Yes, I'm well aware it's big. I've already written some toy compositors to play around with, and have even written an embedded compositor library.
But it's worth doing here. And wlroots does so much heavy lifting that it becomes a lot more manageable. Most of the code in xfwm4 will still be window management code, as it should be.
I've worked on Xfce on and off for 20 years (mostly off, I suppose; probably a total of ~7 years in that time), so I get what these big things are like, and what maintainership means. We're also a very small team of volunteers who have other jobs and other things going on in our lives. But that's fine; Xfce does just fine with that level of work, and will continue to do just fine.
Not sure what you mean. Hopefully I'll be basing it on a released version of wlroots (or at least a non-released version that that will be released by the time I'm ready), which will be dynamically linked just like everything else.
And even if it isn't, and it's statically linked, so what? RAM is cheap and plentiful in most places these days, and if you're really upset about it you can build it yourself with --disable-wayland or whatever.
Anyway, the negativity is quite unnecessary and tiresome.
Because putting everything in shared libs is what made the mess of the bazillion of x11 libs, even though x11 is a IPC protocol.
If I am upset about something, it is wlroots does not manage properly its symbol namespaces with the C preprocessor, that to reduce to near 0 the possibility to have symbol collision while static linking.
The great thing is finally xfce is going to have its wayland compositor. And there, there is not negativity, only positivity, because the real core of wayland is orders of magnitude simpler than x11 core, and wayland is getting its way on mobile, which means with vulkan3d, dude, the future looks bright, not grim, cheer up.
Genuinely find XFCE the only pragmatic option when it comes to "no-fiddling-just-working-and-lightweight" WMs in Linux/BSD land - it's basically like a better Windows 2000 UI with sane defaults throughout. Thanks so much to the devs involved in keeping XFCE up to date. This is major good news.
I love XFCE but nowadays it’s unbearable on a HiDPI screen.
And so I discovered that Cinnamon is also a very good option when it comes to « a better Windows 2000 ui ».
It’s also not incredible on HiDPI but at least Cinnamon is pretty good at not breaking everything when you increase font size by 150 or 200%. On XFCE if you do this everything becomes ugly.
I actually prefer XFCE's scaling via setting a font DPI to e.g. KDE/Gnome's scaling which increases everything in size and not just fonts.
First because usually it is only text that I want to be larger so I can read it. Increases the size of "everything" just decreases UI density with no benefit (while things that conform to the text like button still increase in size to contain the text).
Second because scaling "everything" often leads to ugly results. E.g. I use a program to browse local media files that generates thumbnails for those files. The size of the thumbnail generated matches the size of the widget it displays for that file. If the widget is scaled 1.5x due to UI scaling, it will show a blurry upscaled thumbnail.
Linux Mint XFCE. Appearance > Fonts > DPI Custom DPI setting: 144 makes fonts across the desktop larger. The default setting: 96 seems fine to me. The default in Firefox was -1.0 which is also fine.
I'm using that cheapo LG 24UD58 :) unfortunately they only make 27" now. I really wish they would start making them again. 27" @ 4K is a little bit too low res for me.
Alright I do use KDE those days. I have used XFCE before. I don't think I can believe that KDE is lighter in term of response/lag. I didn't check the ram usage because Firefox is taking it all anyways.
I think KDE can be lighter in terms of response/lag if you turn off animations. On most desktops with dedicated graphics though it probably doesn't matter. Definitely on those older Intel 6000 type graphics and earlier though it does make a big difference.
> And, no Client Side Decorations. XFCE died for me the day they adopted it. Inconsistency-hell ensued.
Not sure what you mean. We have CSDs available for people who like it, but they can be turned on and off.
If you're talking about random GTK apps that run on Xfce, there's not much we can do about that; we don't set app policy.
That's one of the nicer things about Wayland, though: there's a way for the compositor to tell applications "I'll draw the decorations". On X11 there's no protocol for that, only a way for an app to tell the window manager it wants to draw its own, without a way for the WM to say "no, don't do it".
It's absurd that people who recommend "light" Linux distros mention DEs that are much heavier than LXQt and LXDE.
As for Wayland, I use it on a Raspberry Pi 4 (Raspbian xwayland). It works well enough, but there is some kind of bug that causes xwayland to fill the swap space at almost 100%. When this happens, I find it necessary to close applications to reduce the swap usage.
Lighter I won't entirely disagree with: I think a default install of KDE will still be heavier than Xfce, but GTK3 isn't the lightest toolkit, and you can slim KDE down without losing much functionality.
But I don't get the HiDPI point? Certainly 4.16 had issues with HiDPI, but in 4.18 you can set the UI scaling factor, and I believe shortly before and/or after initial release, we fixed up all the issues of blurry image rendering. I have a HiDPI screen and it's worked very well for me, and looks nice.
It used to use more resources and was slower than Gnome 2, when it was still a thing. I don’t know why it got a reputation of being lightweight. I guess people assume less features mean better optimized, which is jot really true. I used to run it for years though, really liked it for a while.
I think it's because Xubuntu was a lighter alternative to Ubuntu when it ran Unity desktop. I don't know how Gnome compares to Unity, but at the time Xubuntu was far better for low-end computers than vanilla Ubuntu so the reputation of XFCE was justified.
this statement is not correct. kde can do both x11 and wayland. checkout the arch-wiki (kde plasma wiki: https://wiki.archlinux.org/title/KDE) to see how.
I've been using it for years and years. I started when it was the only lag free WM I could run on an atom powered netbook (the only "pc" I had at the time)!
Now many years later im still using it, and it still makes me happy. I'm not sure if people know what it's like to be able to click on something and have it work instantly with no lag.
It is a great piece of software, thank you to all the Devs who work on it!
Edit: I've seen a few complaints about hidpi. I'm not sure if this helps, but I have a 4k laptop screen, usb-c to a 4k monitor, and hdmi to a regular 1024 monitor orientated vertical.
By setting x2 in the appearance settings both the 4k monitors look beautiful, and then in the screen manager (where you can move around the screens and set orientation, I just have to set 2x scale on the 1024 monitor and it's fine. Like obviously less good than the 4ks... But it is a worse monitor. I'm surprised it all works so well to be honest!
XFCE being stuck on X11 was what finally made me finally move to KDE when I installed Linux on my machine at work a few months ago
When compiling code on a remote machine, a terminal on one window would cause YouTube videos to lag on the other! As far as I could ascertain, there was simply no hardware accelerated playback going on. Both Chrome and Firefox had seemingly abandoned X11 for a Vulkan based render backend on Wayland
After leaving KDE 5 to "marinate" overnight (I use Gentoo, so compiling an entire desktop environment like KDE can be time consuming) everything "just worked"
I miss the visual simplicity of XFCE and I'm hopeful to return to it one day, but in the interim, responsiveness is key
Firefox works fine on X11 without that setting, but perf ricers will always find something to tweak and complain about on any platform. You can bet some Windows users of Firefox have settings that Firefox is ""unusable"" without too. A certain segment of the population will apply this sort of audiophile mentality to anything you can think of. There are people who think cars are "undrivable" without some fancy hacked ECU, people who think the rental skis at resorts are unusable, and of course people who think the earbuds that came with their phone are completely unlistenable.
Reality is most people aren't sensitive to whatever it is these people believe they are perceiving and will get on fine with whatever the defaults are.
I'm surprised people are saying this because Firefox works perfectly fine on X11 for me without config tweaks
If your hardware is supported (e.g. WiFi chips and GPUs are definitely a valid concern), I know of no post-install config that Debian with Cinnamon needs that you'd not also need to do on Windows, and Microsoft will put ads in your start menu for the trouble of buying their license at that
This guy was using Gentoo on what I'm guessing is an older laptop. That kind of comes with the territory. You don't use LFS or Gentoo as a daily driver and expect not to do some footwork... but if you use Mint, Fedora, Pop_OS, Ubuntu/Debian etc... it pretty much is "just works." Of course, you have to do stuff on pretty much any PC, Windows, Mac or Linux. I own them all and know this to be the case... it's just that some people are blind to the things that they have to do everyday on their OS of choice.
"but if you use Mint, Fedora, Pop_OS, Ubuntu/Debian etc... it pretty much is "just works." "
I just started to make the transfer to Linux from windows. I have limited experience after running Ubuntu on a dual boot laptop back in school a long time ago. It wasnt my daily driver at the time but class required it.
I recently downloaded fedora onto a desktop and it has been a horrid experience. It's sooo slow compared to when windows was on the same hardware. It's so bad the kids won't use the computer unless it's the last one available. I regret the switch, but want the results. I cannot seem figure out what's causing it.
If you want "just works" your more likely to get it with Ubuntu than Fedora. As for being unbearably slow .... there aren't a lot of ways for that to happen unless you're using absolutely the wrong graphics driver.
A typical way would be using xfce's "enable display compositing" setting with a graphics driver that doesn't support proper acceleration.
Using Gentoo as an example of how Linux is hard to use is a bad argument. I have Mac, Linux and windows PCs and they all have fairly ridiculous thing that users of those OS's excuse because they are used to it. I'm not sure how that is an excuse. I'm just reiterating what my og comment said here.
Eh the thing for me is that the config is distributed and fairly easy to correct. If you're fiddling with Windows registry, you better make sure to do a snapshot/backup beforehand. I don't have to do any of that with linux, I just do a simple cp to file.backup and if I hose the system do an emergency boot or linux on usb and fix it.
Yeah, it's pretty much the same story for me; after having found out how to get all the same features in KDE that I had in xfce, there was no way back. I always like to use the lightweight and less resource intensive software, but arguably not using hardware acceleration is actually very resource hungry.
> I always like to use the lightweight and less resource intensive software, but arguably not using hardware acceleration is actually very resource hungry.
What hardware are you using where KDE is considered "resource intensive", yet it also has a GPU that supports Vulkan and hardware video decoding?
I've used KDE for a long time, and can't remember a time when I noticed its resource usage (and I used to daily drive a ThinkPad X200 as late as 2016)
I notice KDE's resource utilization more or less constantly. With KDE's bling turned all the way down, Kwin constantly uses 5-10% CPU when I am literally doing nothing (with my hands off the mouse/keyboard). This causes my laptop fans to noticeably spin up and makes them a lot more prone to spinning up more forcefully while doing a small task. With XFCE, my fan never audibly spins up unless I'm actively doing something. Even if I just do something small/quick, my fans don't spin up most of the time.
I once noticed a memory leak in KDE when setting the background to a (very many image) slideshow instead of a single image. That was probably sometime around 2015.
Normally it's a bit tricky to calculate memory blame between X11, the WM, and the shell, since often their allocations are actually on behalf of an application, and killing the application will reclaim it.
KDE 6 isn't really that much heavier than a "full install" of xfce to be honest. I just prefer the simplicity (probably 1/10th the adjustability of KDE) of xfce for ease of maintenance, but KDE is fine too.
I mained dwm some years ago and I like it, but if I just need a quick environment I usually go back to Xfce. My family's computers (my dad's, my mom's, my sister's and my media PC, because I was lazy) are all Xfce. I will consider switching to Wayland when Xfce will support it.
I still remember seeing a video of Xfce (I think it was version 4.0.4) and being amazed how simple and reliable you could pick your own launchers and widgets on the bar. For me it was much simpler than Gnome at the time.
Modern KDE looks sometimes like Xfce for me (just my feeling, don't bite me), but somehow I always get a lot of strange bugs when I try to change things. Xfce had its own problems (I remember when Thunar was quite unstable), but it still is my go to desktop for simplicity and reliability.
>My family's computers (my dad's, my mom's, my sister's and my media PC, because I was lazy) are all Xfce.
same.
why?
they all grew up on Win9x and the visual metaphors used by XFCE just click, they could all care less what's behind the scenes as long as YouTube and Facebook work.
Yep tho kind of limited in configurability, it's really easy to modify the look of KDE for even casual users. I do wish the would at least ditch the default start menu and use whiskers or something similar though.
In all honesty, the screen sharing situation (and other stuff) in Wayland is currently so brittle that I consider XFCE (my daily driver) lagging behind in its adoption a feature.
I'm curious where you're having trouble. It seemed to me that this was quite bad at first (completely broken) and every application had to add explicit support, but I've had no problem with Firefox, Zoom and Slack recently. Just Microsoft Teams, but webcams don't even work in Teams.
Not the GP, but I have a workflow that depends on the functionality in x11vnc, which allows me to script sharing of individual windows, regions, or desktops. AFAIK, nothing comparable exists in Wayland, and there isn't even the possibility to bring something like that to non-wlroots environments. So I'm stuck on X11 for now.
There's really nothing that can be said then but I personally use `waypipe` like this `waypipe ssh user@127.0.0.1 sway` so that I can run remote sessions and it works amazingly. I sometimes use it to run GUI apps like thunar on my home laptop with tailscale when Im not at home. its basically `ssh -x`
Check out waypipe. It's not compatible with every piece of software, but when it works, it's like one of those magic "run x application over the network" legends except it actually works well.
wayland screen sharing has been solved like 4 years ago and would work already at the time with workarounds.
I remember when we got confined and were sent home, I could not screen share in MS teams on chrome out of the box but a few minutes of search gave me the chrome flags to enable and I just created a new local desktop file to start it up with the right flags
Since then the only things that have been missing is HDR support and color calibration. It sucks to not have them ready but they are only really used by a tiny fraction of the users so for 99.9% of the users wayland works fine.
It is, but with a whole slew of tweaks to pull it off. It's a mess.
...I get it that Wayland is the future and all, that it has features we need as an OS to get to the next level, but I feel there should be no rush to do it, and I feel a strong push for Wayland in distros right now, even with all these little broken things all over the place.
The transition to Wayland also seems to correlate with the adoption of client side decorations (CSD). This "modern" approach destroys the traditional UX of XFCE as seen by recent changes in the settings manager. I fear for the future of XFCE. The advantage of XFCE for me has always been that it's a stable implementation of a traditional Win98/XP UX. I hope they don't adopt more Gnome3 patterns.
> The transition to Wayland also seems to correlate with the adoption of client side decorations (CSD)
Not really. The GNOME/GTK folks were already on the CSD bandwagon well before Wayland. Wayland compositors are free to draw their own decorations (and xfwm4 will indeed continue doing that once it's a Wayland compositor), and one of the actually neat things about Wayland is that there is a protocol that allows the compositor to tell applications not to draw their own decorations. (Whereas on X11 an app can tell the WM it will draw CSDs and the WM can't do a thing about it.)
Certainly GNOME has gone all the way to CSDs (IIRC if an app on GNOME doesn't draw CSDs, they get no decorations at all), but that has nothing to do with Wayland.
What has this to do with wayland or XFCE? The only major DE that is using client side decorations is gnome. This has been true since 2018:
> I heard that GNOME is currently trying to lobby for all applications implementing CSD. One of the arguments seems to be that CSD is a must on Wayland. That’s of course not the case. Nothing in Wayland enforces CSD. Wayland itself is as ignorant about this as X11. [...] In fact we created a protocol (supported by GTK) that allows to negotiate with the Wayland compositor whether to use CSD or SSD.
On Wayland, the compositor can either let windows draw their own borders, or it can disallow it. On X11, if a window want's to draw its own decorations... it will just do it, and you won't be able to do anything about it.
Pretty much every single Wayland compositor follows this except for GNOME, who refuse to do so. It causes problems with windows like Davinci Resolve who don't draw their own decos. This leaves some windows without any controls at all lol. But this is a tangent and purely a GNOME problem.
I mean yeah... But most people are going to have trouble with that, especially being unable to resize or move the window. I even saw a tech Youtuber try out gnome and they were baffled on why they couldn't move or resize davinci resolve whatsoever. Its almost shameful how negligent and hard-headed gnome is being.
Sure it does. Allowing windows to draw their own decorations, often in different ways and with different styles and themes, and not respecting the settings in xfwm4 as to what window-control buttons should be drawn (and where)... that's a huge UX issue.
The process drawing it's own decorations? Technical decision that both Windows (tad complicated) and macOS already do, nothing to do with UX. There's nothing stopping this model from having the uniformity of SSD (as like on macOS), but it does on Linux due to multiple toolkits which don't all agree with eachother.
My accessibility needs are dead under Wayland. There needs to be shortcuts that can be emulated per application and global shortcuts.
Their security model won't allow this. I don't know how accessibility and automation tools are going to get by trying to support every wayland compositor.
It is probably much better if the compositor implements parts of the accessibility functionality. Then you can have other parts as libraries and in the toolkits, but the integration should probably be in the compositor.
If accessibility is a serious concern, Linux is probably not the OS for you. It's so far behind in accessibility compared to Windows it's not even funny.
> If accessibility is a serious concern, Linux is probably not the OS for you. It's so far behind in accessibility compared to Windows it's not even funny.
Linux is fine with X11 per my requirements above. The biggest problem is the fragmentation caused by Wayland. I cant develop software that targets any reasonable range of desktop environments much less, the features needed. If Wayland was left and X11 was right the community needs something in the middle paradigm.
> I cant develop software that targets any reasonable range of desktop environments
I highly recommend Qt for this purpose.
Gtk has been a hot mess for the past few years with Gtk4 and libadwaita shenanigans. However, Qt has continued to evolve incrementally without dropping any backwards functionality.
It's incredibly full-featured and has amazing tooling. In addition to Qt Creator you also have the CLI tools for packaging and whatnot and KDE has Kirigami for components.
The nice thing is that Qt applications look pretty good in native environments like plasma, but they also look pretty good in Gtk environments, including older Gtk-3 desktops and newer Gtk-4 desktops. So you cover all your bases. You also get Mobile and Windows/Mac, if anybody cares.
The two real downsides are:
- GPL
- C++ only (ish). There's other bindings but they will not have the same level and quality of tooling. Gtk has less tooling overall, but MUCH more high-quality bindings.
> > I cant develop software that targets any reasonable range of desktop environments
>
> I highly recommend Qt for this purpose.
I really appreciate your comment. QT is a great framework doing a lot of heavy lifting. However, the accessibility software I help maintain needs the ability to know what windows are in the foreground, all applications running and emulating key/text/mouse input in any of those running applications. Wayland is antithetical to those features by design.
I can see your point with QT for standalone applications. However, with interacting with other applications I'm not sure QT can solve the issue.
If you target just one Wayland compositor, I think this is possible. I mean, each Wayland compositor is basically a display server. The trouble is the Wayland protocol itself doesn't necessarily provide all the tools for this (it does provide some, particularly in conjunction with systems like pipewire and portals).
But, for example, I think kwin can definitely do this.
Yes, I think the design of GTK and GObject worked out really well in the long run. On the surface it seems stupid to use C and then just "bolt on" a C++-like system on top. Why not just use C++? But clearly it's worked out.
Do I understand correctly that they are building on top of wlroots? If so, I am pretty happy about this development as I might get to enjoy the solidity of XFCE apps in other WMs / compositors like Sway or River.
Yes, eventually anyway. I've done a little work on xfwm4, using wlroots, but it'll be another year or more before that's usable. For the desktop components that work on Wayland, we've mostly tested on wlroots-based compositors.
If I had to use a Desktop environment in Linux, it would be XFCE.
But, this Wayland thing is pushing me over to the BSDs. So far I have no abandoned Linux and I doubt Wayland will become mandatory until Firefox stops supporting X.
I sympathize, but unfortunately Tim Cook's stewardship of the Mac, from the embrace of soldered RAM and even storage throughout the Mac lineup, to the increasing annoyances of macOS either in the name of security (e.g., notarization, the popups that request permission to read user directories, which is really annoying when using LLVM's debugger in modern macOS) or to help advertise Apple's subscription services, have harmed the Mac experience. The issue of soldered RAM and storage in particular pushed me back to Windows after more than 15 years of using Macs (which is a shame since Apple's ARM hardware is awesome when it comes to performance and energy consumption); the last time I used Windows as my daily-driver OS was during the XP era. Unfortunately I don't like any of the mainstream desktop operating systems; Windows is annoying and loves to get in the way, macOS is becoming increasingly annoying with each release and is also tied to hardware with massive prices for RAM and storage upgrades with no DIY options like in the past, and the Linux desktop has a Sisyphean development cycle with so much churn. Instead of working on refining the user experience and polishing applications like LibreOffice to make them more competitive to Microsoft Office, it seems like every few years there are major infrastructure changes that require major code changes or even rewrites.
I'm quite disillusioned with desktop computing in general. Microsoft and Google are leading the way with "enshittification" (https://en.wikipedia.org/wiki/Enshittification), and the Mac experience in 2024 isn't like what it was in 2009 when the Mac was at its peak, in my opinion. The Linux desktop ecosystem suffers from a lack of funding relative to its commercial peers, combined with the "organization from chaos" that comes with bazaar-style development. The X11 to Wayland transition would be easier to pull off if the Linux desktop has Microsoft- or Apple-levels of funding and cooperation, but such funding doesn't exist, and part of the characteristic of the bazaar is that developers and teams of developers get to work on projects of their own interest and its the community that "votes" on projects by using them. The bazaar definitely exhibits freedom, but the major downside is that major changes in infrastructure (such as the X11 to Wayland transition) could take a lot of time and have major pain points (for example, the development of incompatible Wayland compositors and the fact that the X11 concept of a window manager doesn't exactly translate to Wayland's concepts).
I love free, open source software, but I'm drawn more to cathedral-style projects such as the BSDs and HaikuOS where there is a coherent vision for the system. Unfortunately the BSDs are dependent on the broader FOSS ecosystem for desktops, which is dominated by the GNOME/KDE/Wayland ecosystem (which also priorities Linux and treats other Unix-like operating systems as an afterthought, but that's another discussion). I'll need to re-investigate HaikuOS to see whether it can be used as a daily driver today; I've seen that a lot of attention has been given to HaikuOS in recent years.
I'm slowly working on a side project where I'm implementing my dream OS, which is influenced by various alternate paths of workstation and desktop computing explored in the 1980s and 1990s but were overtaken by Windows and Unix, namely Lisp machines, the Smalltalk environment, and 1990s Apple initiatives such as OpenDoc and the original plans for the Newton, which involved a Lisp OS (https://mikelevins.github.io/posts/2021-07-12-reimagining-ba...). But I have a busy day job and so this will take years of effort on the side.
I'll be horribly blunt here but I don't give a shit about any of that any more. It's irrelevant which is why open source is a complete joke for most people.
I, much as other users, just want my computer to solve the problems I ask of it, not turn into an activity in its own right or a maze of hosting problems and pain to get a simple task done. The job of the computer is to free us from slavery and everyone seems to have bloody well forgotten that.
macOS is my dream OS as it stands.
I can add a cucumber to my shopping list on my mac and when I get to the supermarket, it's on my phone.
I can share a calendar with my family and it just works.
I can listen to music on it and the same music is on my phone.
I can open and move pages around in a PDF without having to futz with anything or open a terminal.
I can literally write on my iPad with a pen and a PDF pops out full of my writing which I can send to people for review and feedback.
I can use Lightroom, Photoshop and Excel, which are far superior to any open source products and I do not mind paying for that privilege.
etc etc.
The hardware cost and upgrade inflexibility, neither of which I have found a problem either way, is insignificant anyway to the pain I would incur without it.
And that's what the community does't get: the last 20% is the hard bit and the above is the last 20% and it has never even once in the history of OSS got anywhere even remotely near that. No one cares about the political purity and ideology behind it.
> And that's what the community does't get: the last 20% is the hard bit
I think a lot of people in OSS do get that, but in many ways it's significantly harder to do that last 20% in an open source project than it is for a proprietary software company to do it.
But ultimately I don't care that much, honestly. If you don't get what you need out of open source, that's fine. You not using it isn't a problem for me or for my own open source work. As long as there are enough users to keep communities going and developers interested, that's all that matters.
> and the above is the last 20% and it has never even once in the history of OSS got anywhere even remotely near that.
That's provably false, and it's pretty annoying when people make uninformed absolutist statements like that.
> No one cares about the political purity and ideology behind it.
A lot of people do. No, not a majority, and maybe not even a largely sizeable minority, but it's very very far from "no one". If you don't, I guess that's fine. I hope your proprietary software continues to do what you need and want, and that your needs and interests continue to be aligned with what those software companies want to give you. That hasn't been true for me for a very long time now, but certainly people have different needs and different thresholds for it to matter. But if that changes, feel fortunate that all the work we've done will still be available for you to take advantage of.
> The job of the computer is to free us from slavery and everyone seems to have bloody well forgotten that.
Your use of the word "slavery" is a bit bizarre and overwrought, but it's funny you say something like this. Open source software is the only kind of software that's going to respect your freedom. Proprietary software never will. You're at the whims of its makers, and can never take control of your own computing destiny.
Like the sibling said, I guess that doesn't matter to you now. But maybe it will, someday, when the game changes yet again.
As I've said elsewhere, your data is the important bit. And at least in my case that is 100% portable. The winds of change are inevitable, but you can move house without too much hassle.
If think you can trust your data with a bunch of proprietary software companies, I'm not sure what to tell you.
> And at least in my case that is 100% portable
For now. Time and time again I see stories where people think they have redundant copies of everything important, and can reconstruct things when a company pulls out the rug out from under them, only to find they're missing something important.
Maybe you aren't. Great, good for you. You represent a teeny tiny percentage of the sum of proprietary software users. Again, I'm glad you believe you're safe and happy with the arrangement you have, but you're uncommon, and that's the entire point.
And for me, it's not just portability. Most companies are going to mine your data, use it to train AI, sell it to third parties, whatever they think will make them money. Or they just get hacked, like everyone does, eventually. You can protect yourself from that with open source. Not saying you don't still have to be vigilant, of course you do... but with proprietary software? Well, best of luck with that. Even with Apple, one of the few companies I think are reasonably trustworthy stewards of customer data, despite all the other issues I have with that company.
> No one cares about the political purity and ideology behind it.
No one cares, until they do. Everything you mentioned depends on a single company, which follows the commands from just a few people, who can one day decide to lead the operating system in a direction which makes it no longer your "dream OS", and intellectual property law makes it so that no other company can undo their changes.
> The job of the computer is to free us from slavery and everyone seems to have bloody well forgotten that.
We haven't, and that's why we prefer operating systems we fully control, instead of being bound by the whims of some foreign megacorporation, enforced by intellectual property law.
The other side of the fence is just as bad. Look at what Redhat and Freedesktop did to Linux.
While you might have the illusion of control, the exit plan to take control is to take on tens of millions of lines of source code and manage it yourself. That is not control.
Your data is the important bit. If you keep that portable and/or recreatable then it doesn't matter what tools you use.
> Look at what Redhat and Freedesktop did to Linux.
Well now you're just contradicting yourself. I'm not going to say RedHat and fd.o have been perfect all the time or have always made the best decisions or haven't pursued agendas that may have been at odds with others.
But c'mon. On one side you're saying that open source can never be good because no one maintains the cohesiveness and attention span to do "the last 20%", but then you rail on organizations that have tried to set and maintain standards, and reduce fragmentation on the desktop?
> I'm slowly working on a side project where I'm implementing my dream OS, which is influenced by various alternate paths of workstation and desktop computing explored in the 1980s and 1990s but were overtaken by Windows and Unix, namely Lisp machines, the Smalltalk environment, and 1990s Apple initiatives such as OpenDoc and the original plans for the Newton
Cool, I'll be following that. To me, one of the most interesting aspects of Genera is the command line interface, where commands know their arguments' names, types, defaults, and help strings. The commands can output nicely formatted data such as tables and, because they're just Lisp functions, they can share structured data directly rather than parsing and serializing ad-hoc text formats as is that's typical in the Unix world.
I've never used Genera; I was born in 1989 and thus was too young to experience Lisp machines. However, I've read a lot about them over the years, and I wish we had a modern analogue. I think the closest thing we have to Genera is the Squeak implementation of Smalltalk and its derivatives (Pharo and Cuis Smalltalk).
I wholeheartedly agree with you; a system based on structured data or objects would be much better than what we have today. What I've been dreaming about for years is a system where everything is an object, much like Smalltalk. Instead of typing "ls" in a terminal window, you would send a get_files request to the current directory, which would return a list of file names. It's possible to emulate the traditional Unix "ls" command, but interacting programmatically with the system enables so many other uses that's much easier than parsing text or calling POSIX system calls. Because the user is working with objects instead of text, not only would it be easier to write scripts compared to Unix shell scripts, it would also be easier for developers to implement GUI tools for the system. For example, imagine if word processing documents were represented by a class that handled all sorts of manipulations of the document, hiding the details of the underlying file format from the user and developer in most cases. It would be easier to come up with scripts for generating and modifying word processing documents without having to use the word processor's macros (if they're available in the first place) or without having to dive deep into the underlying file formats of the documents.
Oh, I'm also too young to have experienced the Lisp hardware. I've tried Genera in an emulator, though, and I've read some of the manuals because it's so interesting. There's a nice user interface programming manual preserved at [1]. Defining commands starts on PDF page 39.
I do not like the panels Apple uses, they have harsh backlight, due to 450 nm blue primary; nwer panels have 455-460nm blue light peak. My eyes are sensitive and I get tired on Apples.
Have you tried blue-block type lenses? Not those yellow ones from infomercial - it's a clear lens that cuts some wavelength for staring at bright monitor all day. Lots of cheap readers have those.
This was not my point. My point is that Apple uses old panels, which are pushier on eyes, instead of asking LG to put normal backlight. A supposedly premium manafacturer stuck with the panel from 2016; Apple stopped innovating, outside the M cpu series.
Btw the lenses still are yellowish, and interfere with white balance. Attempt to correct the white balance will destroy the effect of the lenses at 450 nm will be back.
Soldered RAM is faster than socketed RAM. What's faster still is Apple's current approach: putting the RAM inside the CPU chip package. This is one of the reasons why Apple Silicon chips are so screamingly fast and handily beat any x86 chip core-for-core.
> to the increasing annoyances of macOS either in the name of security (e.g., notarization, the popups that request permission to read user directories, which is really annoying when using LLVM's debugger in modern macOS)
These help prevent malicious software, such as downloaded apps or even web pages, from exfiltrating or tampering with user data. Relevant xkcd: https://xkcd.com/1200/
X11 is basically already unsupported, pretty much all the developers moved to Wayland. Even the BSD and everybody else relaying on X11 has noticed that and they know they can't maintain X11 by themselves.
It will 100x not go away, there is no way. Literally all big stakeholder both open source (Gnome, KDE, Fedora and so on) and companies like Valve are all in on Wayland.
The reality is, X11 is a complete train wreck and always was. Its just that for a long time people invest in it and made it usable. The same is true for Wayland, things have been improving rapidly in Wayland and it many ways its already better.
The train wreck is being cleared and the trains are running again, literally nobody will go back to horses.
When software works it just works, and if you don't change it it continues to work. But if you had an issue, its unlikely that it would be fixed or that new features you want would arrive. Unsupported is fine for much software on the world.
Hardware is evolving, and while the core feature set works, it wont keep up. If you personally don't care about those things, that fine, and its a feature for you. But eternal stagnation isn't actually a realistic option.
I mean yeah. Wayland started with a feature aversion from its initial developer, then continued with feature avoidance by committee, the committee in turn required by the "everyone gets to implement the X server equivalent" design and some of the trouble coming particularly from "everyone" including Gnome. Some of these missing features are really not considered optional at this point by many desktop Linux users. But there aren't many left now. I'll probably make the switch in 0.5-3 years ;)
What is there in Wayland is really quite clean though, so it will... EVENTUALLY... be pretty good.
>X11 is basically already unsupported, pretty much all the developers moved to Wayland
In my view, Wayland is basically unsupported. GNOME and KDE support it. XFCE, Lxqt, Mate, and Cinnamon don't. It's been over 10 years. You could make a whole DE from scratch in this time.
For most of those years, fundamental parts of the desktop experience were just not possible on Wayland. Even today, there are some current Xfce features that are just not implementable on Wayland.
> You could make a whole DE from scratch in this time.
Depends on your goals and team size. Part of me would love to rewrite Xfce in Rust (yes, I'm one of those people), using a different, lighter, actually-well supported toolkit (unlike GTK3), but with the team we have, that would be another two decades.
Consider that Xfce 4 is a little over 20 years old at this point. It's taken that long to build it to what it is. Starting over wouldn't take any less time to get there with the people and resources we have.
I don't agree, because experimental support means it's being worked on. Legacy support means the opposite - it's not being worked on and never will be worked on.
So for now it happens to work. But none of those DEs are contributing to the X.org implementation of X11, right? So it will slowly break down over time. That's the current status of X11, it's on hospice.
>Wayland has a myriad of unresolved problems regarding surface suspension blocking presentation and the FIFO (vsync) implementation being fundamentally broken leading to reduced GPU-bound performance.
To which someone replies:
>If we do this, we are basically accepting these issues are unfixable for the next ten years (SDL4). Having this as the default in SDL3 (which isn't being used yet!) is important for signaling to other stakeholders that we actually do need to get solutions in place for detected issues.
To which the committer says:
>SDL is not your tool for "signaling to stakeholders" about what is important. It's an actual library used by real developers and users!
Likewise. My PC is not a tool for signaling library/DE developers that there are more and more people using Wayland now and that they can finally stop maintaining X11 legacy code. I use my PC for real things. I couldn't care less if things are running on X11 or Wayland so long as they work.
If Wayland isn't working with my use case, I just won't use it. I won't switch to Wayland juts to beta test it, and then spend years begging people who know how to hack Linux to implement the features I need that I had on X11.
If Linux drops X11, and Wayland still doesn't work. I won't use Wayland. I'll use Windows.
> they can finally stop maintaining X11 legacy code
They already have, this is what I'm trying to tell you. All the X.org maintainers are on Wayland, and none of the DEs are contributed to X.org. It's on hospice. It happens to work just because, but that won't be the case forever.
It's fine if you currently like X and use X. I myself use X on Debian Stable.
But the reality is that it's dead technology, and eventually it won't work right.
Wayland is constantly being improved by many, many different parties. Valve, KDE, Gnome, freedesktop, Redhat... they're all constantly making new protocols and implementing them to solve problems.
They haven't stopped maintaining it because the programs still run on X11.
GIMP supports PNG. That doesn't mean GIMP contributes code to libpng, or libpng needs updates. It just needs to have code to make it work with PNG.
X11 will stop being supported when the developers literally remove the code to make their programs run on X11 from their projects. Until then, it's still supported.
Why would I need updates when things already work in the current version?
All X11 programs already run under Wayland. There will always be applications that are X11-only and they will always work. This is a fake problem, IMO.
> Why would I need updates when things already work in the current version?
I guess you don't, but then again, I don't see you running Windows 95. I think this is maybe something you like to believe, but in practice people do want new things and new software.
You maybe don't want to hear it, but the overwhelming majority of people are on Gnome or KDE. You might as well claim X11 is not supported because Hyperland, Cosmic and many others aren't supported on X11.
Unsupported in the context of software doesn't really mean what you took it as anyway.
"Unsupported" as in there are no maintainers, there's no roadmap, there's no future. X11 is a dead end. All the people who were working on it would rather work on Wayland instead. It's taking a while for the DEs, etc. to catch up, but even the slow pace of improvement Wayland is exhibiting is much better than "dead in the water" which is where X11 is.
And so is X11. We are heading for a dual stack desktops. There are tons of expensive applications that target Linux _because_ of X11 and they will rather force users to switch OSes than get rewritten.
Pretty much any app that uses a toolkit won't need to be rewritten. Most of them just need small changes, or even no changes.
Those that are written to use Xlib directly, sure, they'll take some work (though rewritten? doubtful). If their maintainers don't want to do that, they'll still run under Xwayland.
These apps were written for X11 and NFS, so they can run on any node in a cluster. The were typically deployed on commercial UNIX systems and ported to Linux when Sun went belly up. In practice, I don't think porting them again will be needed. It is enough to qualify a different distribution if Redhat chose to exit the workstation market. There are already companies like NoMachine that keep X11 running.
It is not even about the access (people usually use nx or even vnc for that). It is about having a system that can work as a part of a cluster (network filesystem, X11), with all base GUI utilities (terminals, file managers, browsers, pdf viewers) supporting that.
> And so is X11. We are heading for a dual stack desktops.
IMO, we're heading for "X11 over Wayland" desktops, in which the only X11 implementation is Xwayland. These expensive applications which target X11 will keep working, through that compatibility layer.
There are other Unix systems but by far the easiest solution is simply making sure all installed apps can still behave as X11 clients. Wayland or Xwayland is not really any issue. Few people use Xorg directly, as the new deployments usually require some form of a remote access.
I haven’t figured out the magic dbus (I guess) incantations required to make screen sharing work without systemd though. I guess it needs pipewire at least.
Perhaps if the X devs want people to migrate, they should consider giving Wayland all the features of X. Otherwise people are going to keep using the thing that works.
No, but there should exist a print server that Wayland applications can use. I don't object to refactors and replacements, but the underlying features should still exist.
There's 0 point in making a new software if you're just going to inherit all the fundamental design flaws that make the original software unmaintainable.
GPUs live in desktops and phones, but apps live in remote datacenters, so the future seems to be sending rendering commands to Javascript frontends on end users' browsers.
Xfce is not perfect but pretty close (in *nix realm); it is a stable WIMP desktop environment. Everything where you expect it to be, no excessive feature creep, stable. It is like a tool, I love it. I wish it had a better brand. Too nerdy.
Most projects introduced bugs in xorg trying to support wayland, which is not reliable for many use cases, making everything unusable every other month
That's the plan, anyway. And I expect Wayland won't be usable for most as a daily driver until 4.22 or even 4.24, so obviously X11 will remain a high priority. And even after then, the X11 code paths will still work. Most of the difference is handled inside GDK anyway; the work we've done for Wayland support is mainly where GDK doesn't handle what we want to do, and where we've had to write libX11 code.
Ok, I’m super confused, how much work is involved in supporting Wayland vs Xserver?
[editing to add: my assumption that it isn’t is due to this post - xfce is what I used to use because it was lightweight, but if such a lightweight manager requires significant adoption effort it would imply that Wayland isn’t X API+new extension]
I had thought that Wayland was still functionally an implementation of the X server protocol, just reengineered to drop any pretense of network transparency, but given the amount of work involved that’s clearly not the case?
(It has been a long time since I did anything resembling X development, and so I’ve paid literally no attention to the details)
The core wayland does not even implement the idea of a window, to show how far the differences go.
Based on experiences of an acquaintance writing from scratch support for wayland for an UI framework that shares absolutely zero code with any of the major ones, the differences in even basic things like how to implement drop-down menu are staggering.
Funnily enough, said acquaintances needs would have been fully covered by slightly more capable XRender and a way to synchronize to VSync that does not depend on GLX.
The oversimplified answer is that a Wayland compositor must include all the functionality of an X window manager and the Xorg display server. There are libraries to help implement all that, but it's still a big lift.
Oh, ok, so X apps can run under Wayland, because Wayland DEs can run an X server (an existing X impl, or does Wayland provide an X translation layer/impl?)
But the DE/window manager is what is actually responsible for deciding when/where to provide that interface? For actual rendering+windows etc the DE is communicating entirely through Wayland specific APIs?
I guess the operation mode is similar to how you would run X apps on Mac and windows where you could launch an x server that would act as a layer between the Mac/windows windowing system?
Wayland is written by most of the same folks who maintain xorg, and XWayland is part of xorg.
It's generally more transparent than running a separate X server on Mac or Windows -- you don't normally need to care about whether an app is X11 or native Wayland.
Xwayland is basically Xorg with a Wayland DDX layer -- and it recently got a rootful mode. So that means it's possible to run a complete X desktop, including WM, on top of a Wayland compositor.
However, the drawback is that with the exception maybe of Weston, one does not simply start up a Wayland compositor. You need to have support for logind seats and that sort of thing. So unless you want to spend an afternoon assembling the components like a piece of flat-pack furniture from Target, you basically need systemd and a distro that bundled those components for you the right way.
Not really. It's perfectly fine to just advertise a single dummy wl_seat, and use whatever non-logind mechanism there is to tell the OS that you want to be in control of the input and display hardware.
You can run any wlroots- (and possibly smithay-) based Wayland compositor nested inside an X11, but it won't be rootless; you'll get a separate window that all the Wayland apps will be drawn into.
There's really no reason why someone couldn't write a "WX11" or something of that nature, but I'm not sure it would be worth the effort.
A desktop environment in X11 is leaning on the X server. One on Wayland has to have libs for most of the things. There is no window manager. A compositor is like a server and a window manager in one. There are libs, but the process is much more involved and the API is much more tight so not everything is easy to get as in X.
Wayland is still broken in so many ways it's not funny.
Just recently I had to switch back to X11 because Wayland was just awful performance wise, like stuttering and constant fps drop. And this on a what 2023 high quality laptop with standard Ubuntu.
Xfce is pronounced “ecks-eff-see-ee”. The name Xfce originally stood for “XForms Common Environment”, but since then Xfce has been rewritten twice and doesn't use the XForms toolkit anymore. The name survived, but it is no longer capitalized as “XFCE” and is no longer an abbreviation for anything (although suggestions have been made, such as “X Freakin' Cool Environment”).
In this case, because XFCE is based on GTK, and the backend for GTK has gotten a lot of support for native Wayland recently. Starting from scratch would probably end up duplicating a lot more code than necessary, whereas supporting both in one codebase helps keep the featureset in lockstep for both x11 and Wayland users.
To be clear here, we don't have a compositor yet. I've done a little work on it, but it won't be usable for at least a year, maybe (probably) longer.
4.20 will let you run some of our components on another compositor. We mostly test with Wayfire and Labwc, as our compositor will also be using wlroots.
Good luck and thanks for all the work. XFCE has been my daily driver for a few years, the kind of tool that is so on point you forget it's there.
XFCE has been my preferred Unix desktop for like 20 years.
Great software. Great mix of flexibility, modularity, ease of use, and good taste.
I also thank the devs.
Before I went all in on i3 then Sway, XFCE was the cats pyjamas. Still is the dogs boxers as far as I am aware.
Awesome! A lot of people are excited, XFCE is a very charming and beloved desktop environment. I'm curious to see how people will take this change though... I know a lot of XFCE users either want Wayland or don't care, and then there are a lot who will probably be very resistant to this change. Taking it slow at least allows you to take advantage of the recent growth of Wayland and allows you all to polish XFCE Wayland up so that there are minimal hiccups during the switch. I'd be curious to know what your opinion is on the matter.
It's not so much a change as it is (well, will be) an option. We're not abandoning X11, and I expect it will be a couple more major releases before we have feature parity on Wayland with what we have in X11.
But yes, some people are excited, some don't care, some are worried.
But there won't be a switch. If users want to keep running it on X11, they can do that. I can't promise that will be the case forever, but it's not (won't be) that hard to support both with how things are currently.
(uninformed outsider) have you considered reaching out to the System76 folks? They are also setting out doing a Wayland compositor for a (decreasingly) gtk-based desktop and are reasonably far along on this goal. I wonder if their compositor would suit your needs?
To me, at least, xfwm4's window management behavior is a big part of what makes Xfce... well, Xfce. So no, I don't think I'd consider using someone else's compositor.
To add a bit, hours later: I've been testing my Wayland support for the other components using Wayfire and Labwc, and honestly I find their window management incredibly frustrating. Part of it is just lack of familiarity, I'm sure, but honestly I just don't like how they do things.
A lot of xfwm4's behavior is windowing-independent, or fairly easy to put the windowing-specific calls behind interfaces. There's some of it that turns out to be pretty hard to untangle, which is where the real work is, but I think it's worthwhile work to take on.
Sure, the xfwm4 behaviour is absolutely something you'd need to be able to own in a technical and organisation sense. However all the other stuff around building a Wayland compositor is... big. Sharing portals, XWayland support, high dpi and mixed low/high dpi, display management as a whole, and that's just the stuff I know about..
I love xfce, I don't use it anymore though because I wanted Wayland support. As a (past and potentially future) user, I want to see a healthy xfce, not one that is burning out its devs with the giant task of maintaining a Wayland compositor. :)
Yes, I'm well aware it's big. I've already written some toy compositors to play around with, and have even written an embedded compositor library.
But it's worth doing here. And wlroots does so much heavy lifting that it becomes a lot more manageable. Most of the code in xfwm4 will still be window management code, as it should be.
I've worked on Xfce on and off for 20 years (mostly off, I suppose; probably a total of ~7 years in that time), so I get what these big things are like, and what maintainership means. We're also a very small team of volunteers who have other jobs and other things going on in our lives. But that's fine; Xfce does just fine with that level of work, and will continue to do just fine.
System76 is using Rust and Iced to create the COSMIC desktop. It is not GTK anymore. That drags in a lot of dependencies for a project like XFCE.
Isn’t compiz a compositor for xfce?
Btw I love xfce. Find it to be the perfect balance between features, stability and performance!!!
Compiz is an X11 compositor that can be used with Xfce if you disable xfwm4's builtin compositor.
A Wayland compositor is a completely different beast, and that's what we have to build.
Until wlroots is statically linked into xfce compositor... you know, if we could avoid the mess of the bazillion of x11 libs all over again...
Not sure what you mean. Hopefully I'll be basing it on a released version of wlroots (or at least a non-released version that that will be released by the time I'm ready), which will be dynamically linked just like everything else.
And even if it isn't, and it's statically linked, so what? RAM is cheap and plentiful in most places these days, and if you're really upset about it you can build it yourself with --disable-wayland or whatever.
Anyway, the negativity is quite unnecessary and tiresome.
Because putting everything in shared libs is what made the mess of the bazillion of x11 libs, even though x11 is a IPC protocol.
If I am upset about something, it is wlroots does not manage properly its symbol namespaces with the C preprocessor, that to reduce to near 0 the possibility to have symbol collision while static linking.
The great thing is finally xfce is going to have its wayland compositor. And there, there is not negativity, only positivity, because the real core of wayland is orders of magnitude simpler than x11 core, and wayland is getting its way on mobile, which means with vulkan3d, dude, the future looks bright, not grim, cheer up.
Genuinely find XFCE the only pragmatic option when it comes to "no-fiddling-just-working-and-lightweight" WMs in Linux/BSD land - it's basically like a better Windows 2000 UI with sane defaults throughout. Thanks so much to the devs involved in keeping XFCE up to date. This is major good news.
I love XFCE but nowadays it’s unbearable on a HiDPI screen.
And so I discovered that Cinnamon is also a very good option when it comes to « a better Windows 2000 ui ».
It’s also not incredible on HiDPI but at least Cinnamon is pretty good at not breaking everything when you increase font size by 150 or 200%. On XFCE if you do this everything becomes ugly.
I actually prefer XFCE's scaling via setting a font DPI to e.g. KDE/Gnome's scaling which increases everything in size and not just fonts.
First because usually it is only text that I want to be larger so I can read it. Increases the size of "everything" just decreases UI density with no benefit (while things that conform to the text like button still increase in size to contain the text).
Second because scaling "everything" often leads to ugly results. E.g. I use a program to browse local media files that generates thumbnails for those files. The size of the thumbnail generated matches the size of the widget it displays for that file. If the widget is scaled 1.5x due to UI scaling, it will show a blurry upscaled thumbnail.
Hmm, can't agree, having run Xfce for years on HiDPI screens.
Go to Settings, Appearance, Advanced, set the font DPI to something like 144 or what works for your screen.
The only thing that does not obey this is Firefox; the way to adjust it is to set layout.css.devPixelsPerPx to something like 1.5 in about:config.
Linux Mint XFCE. Appearance > Fonts > DPI Custom DPI setting: 144 makes fonts across the desktop larger. The default setting: 96 seems fine to me. The default in Firefox was -1.0 which is also fine.
Last time I tried (a few months ago), the text was indeed bigger but most window layouts were broken, the icons kept the same size etc etc …
> I love XFCE but nowadays it’s unbearable on a HiDPI screen.
How so? We support setting GTK's UI scaling to 2x, and (as of 4.18) I believe all the bugs around blurry icon rendering have been fixed.
What exactly becomes ugly for you?
KDE is really excellent on HiDPI. Running it here on a 4K 24" screen and it looks amazing
What monitor do you use ? 24" 4K is such a rare thing these days ; I'm using cheapo LG , but I always look out if there's something better
I'm using that cheapo LG 24UD58 :) unfortunately they only make 27" now. I really wish they would start making them again. 27" @ 4K is a little bit too low res for me.
HiDPI pushed me to run Mint for a while, then multi-monitor issues with hiDPI finally pushed me to WSL.
Been using WSL full time for over a year and have no plans to go back..
KDE is actually lighter. And hidpi is much better on Qt.
Alright I do use KDE those days. I have used XFCE before. I don't think I can believe that KDE is lighter in term of response/lag. I didn't check the ram usage because Firefox is taking it all anyways.
Auto Tab Discard [1] does wonders to Firefox RAM consumption if you open lots of tabs. Has the "recommended" badge.
[1]: https://addons.mozilla.org/en-US/firefox/addon/auto-tab-disc...
Absolutely! It was a game changer for me. No more OOM with 16GB RAM
I think KDE can be lighter in terms of response/lag if you turn off animations. On most desktops with dedicated graphics though it probably doesn't matter. Definitely on those older Intel 6000 type graphics and earlier though it does make a big difference.
And, no Client Side Decorations. XFCE died for me the day they adopted it. Inconsistency-hell ensued.
LXQt if you want something with less features, Plasma if you want more features. They're both performant.
> And, no Client Side Decorations. XFCE died for me the day they adopted it. Inconsistency-hell ensued.
Not sure what you mean. We have CSDs available for people who like it, but they can be turned on and off.
If you're talking about random GTK apps that run on Xfce, there's not much we can do about that; we don't set app policy.
That's one of the nicer things about Wayland, though: there's a way for the compositor to tell applications "I'll draw the decorations". On X11 there's no protocol for that, only a way for an app to tell the window manager it wants to draw its own, without a way for the WM to say "no, don't do it".
> If you're talking about random GTK apps that run on Xfce, there's not much we can do about that; we don't set app policy.
Yes, guess you're correct. Qt apps in Plasma was a better mix for me. I have consistency and no funky titlebars.
It's absurd that people who recommend "light" Linux distros mention DEs that are much heavier than LXQt and LXDE.
As for Wayland, I use it on a Raspberry Pi 4 (Raspbian xwayland). It works well enough, but there is some kind of bug that causes xwayland to fill the swap space at almost 100%. When this happens, I find it necessary to close applications to reduce the swap usage.
What do you mean by heavy though? Plasma is feature heavy but definitely not memory heavy.
Lighter I won't entirely disagree with: I think a default install of KDE will still be heavier than Xfce, but GTK3 isn't the lightest toolkit, and you can slim KDE down without losing much functionality.
But I don't get the HiDPI point? Certainly 4.16 had issues with HiDPI, but in 4.18 you can set the UI scaling factor, and I believe shortly before and/or after initial release, we fixed up all the issues of blurry image rendering. I have a HiDPI screen and it's worked very well for me, and looks nice.
GTK3 had no fractional scaling, GTK4 had it broken, but QT has the best fractional scaling I've seen.
Source?
It used to use more resources and was slower than Gnome 2, when it was still a thing. I don’t know why it got a reputation of being lightweight. I guess people assume less features mean better optimized, which is jot really true. I used to run it for years though, really liked it for a while.
I think it's because Xubuntu was a lighter alternative to Ubuntu when it ran Unity desktop. I don't know how Gnome compares to Unity, but at the time Xubuntu was far better for low-end computers than vanilla Ubuntu so the reputation of XFCE was justified.
Xubuntu was lighter than Ubuntu (when it had Gnome), at least that is how I remember it.
But unfortunately KDE is Wayland only now, so no way to turn the compositor off.
That is just false. Source: the latest openSUSE Tumbleweed packages provide an up-to-date X11 version of kwin.
this statement is not correct. kde can do both x11 and wayland. checkout the arch-wiki (kde plasma wiki: https://wiki.archlinux.org/title/KDE) to see how.
I run x11 on my Debian 12.
I do too, but this doesn't really count because Debian Stable is still plasma 5.
definitely my go to for VM's and VNC for lightweight no need to fiddle installs.
If you really want a just working and actually lightweight (instead of pretending like XFCE), you choose icewm.
apples and oranges tho. you can use icewm instead of XFWM
Xfce all day every (work) day.
I've been using it for years and years. I started when it was the only lag free WM I could run on an atom powered netbook (the only "pc" I had at the time)!
Now many years later im still using it, and it still makes me happy. I'm not sure if people know what it's like to be able to click on something and have it work instantly with no lag.
It is a great piece of software, thank you to all the Devs who work on it!
Edit: I've seen a few complaints about hidpi. I'm not sure if this helps, but I have a 4k laptop screen, usb-c to a 4k monitor, and hdmi to a regular 1024 monitor orientated vertical.
By setting x2 in the appearance settings both the 4k monitors look beautiful, and then in the screen manager (where you can move around the screens and set orientation, I just have to set 2x scale on the 1024 monitor and it's fine. Like obviously less good than the 4ks... But it is a worse monitor. I'm surprised it all works so well to be honest!
XFCE being stuck on X11 was what finally made me finally move to KDE when I installed Linux on my machine at work a few months ago
When compiling code on a remote machine, a terminal on one window would cause YouTube videos to lag on the other! As far as I could ascertain, there was simply no hardware accelerated playback going on. Both Chrome and Firefox had seemingly abandoned X11 for a Vulkan based render backend on Wayland
After leaving KDE 5 to "marinate" overnight (I use Gentoo, so compiling an entire desktop environment like KDE can be time consuming) everything "just worked"
I miss the visual simplicity of XFCE and I'm hopeful to return to it one day, but in the interim, responsiveness is key
Firefox on X11 works fine if you enable media.ffmpeg.vaapi.enabled in about:config
Thanks for the tip. I didn't have it enabled, but to be honest I've never noticed any lag when playing back videos or in any other situation.
With all the cores in modern PC's I'd doubt if you would unless you're on a computer that is 8 or 10 years old
I tried enabling that. I was able to verify vaapi was enabled. But even using a "forceh264" plugin, Youtube defaulted to unaccelerated playback
Y'all need to remember this in those threads when you say Linux just works and MS's registry is stupid and needs tweaking to work right.
Firefox works fine on X11 without that setting, but perf ricers will always find something to tweak and complain about on any platform. You can bet some Windows users of Firefox have settings that Firefox is ""unusable"" without too. A certain segment of the population will apply this sort of audiophile mentality to anything you can think of. There are people who think cars are "undrivable" without some fancy hacked ECU, people who think the rental skis at resorts are unusable, and of course people who think the earbuds that came with their phone are completely unlistenable.
Reality is most people aren't sensitive to whatever it is these people believe they are perceiving and will get on fine with whatever the defaults are.
I'm surprised people are saying this because Firefox works perfectly fine on X11 for me without config tweaks
If your hardware is supported (e.g. WiFi chips and GPUs are definitely a valid concern), I know of no post-install config that Debian with Cinnamon needs that you'd not also need to do on Windows, and Microsoft will put ads in your start menu for the trouble of buying their license at that
This guy was using Gentoo on what I'm guessing is an older laptop. That kind of comes with the territory. You don't use LFS or Gentoo as a daily driver and expect not to do some footwork... but if you use Mint, Fedora, Pop_OS, Ubuntu/Debian etc... it pretty much is "just works." Of course, you have to do stuff on pretty much any PC, Windows, Mac or Linux. I own them all and know this to be the case... it's just that some people are blind to the things that they have to do everyday on their OS of choice.
"but if you use Mint, Fedora, Pop_OS, Ubuntu/Debian etc... it pretty much is "just works." "
I just started to make the transfer to Linux from windows. I have limited experience after running Ubuntu on a dual boot laptop back in school a long time ago. It wasnt my daily driver at the time but class required it.
I recently downloaded fedora onto a desktop and it has been a horrid experience. It's sooo slow compared to when windows was on the same hardware. It's so bad the kids won't use the computer unless it's the last one available. I regret the switch, but want the results. I cannot seem figure out what's causing it.
If you want "just works" your more likely to get it with Ubuntu than Fedora. As for being unbearably slow .... there aren't a lot of ways for that to happen unless you're using absolutely the wrong graphics driver.
A typical way would be using xfce's "enable display compositing" setting with a graphics driver that doesn't support proper acceleration.
Not an older laptop. Some 9th gen Intel i7 machine with a Radeon RX6400 GPU attached
Always an excuse... Why can't you just agree that this isn't how it should be but maybe you like Linux anyway?
Using Gentoo as an example of how Linux is hard to use is a bad argument. I have Mac, Linux and windows PCs and they all have fairly ridiculous thing that users of those OS's excuse because they are used to it. I'm not sure how that is an excuse. I'm just reiterating what my og comment said here.
Eh the thing for me is that the config is distributed and fairly easy to correct. If you're fiddling with Windows registry, you better make sure to do a snapshot/backup beforehand. I don't have to do any of that with linux, I just do a simple cp to file.backup and if I hose the system do an emergency boot or linux on usb and fix it.
It never worked on my (now dead) Asus netbook, although Chrome could do it.
Yeah, it's pretty much the same story for me; after having found out how to get all the same features in KDE that I had in xfce, there was no way back. I always like to use the lightweight and less resource intensive software, but arguably not using hardware acceleration is actually very resource hungry.
> I always like to use the lightweight and less resource intensive software, but arguably not using hardware acceleration is actually very resource hungry.
What hardware are you using where KDE is considered "resource intensive", yet it also has a GPU that supports Vulkan and hardware video decoding?
I've used KDE for a long time, and can't remember a time when I noticed its resource usage (and I used to daily drive a ThinkPad X200 as late as 2016)
I notice KDE's resource utilization more or less constantly. With KDE's bling turned all the way down, Kwin constantly uses 5-10% CPU when I am literally doing nothing (with my hands off the mouse/keyboard). This causes my laptop fans to noticeably spin up and makes them a lot more prone to spinning up more forcefully while doing a small task. With XFCE, my fan never audibly spins up unless I'm actively doing something. Even if I just do something small/quick, my fans don't spin up most of the time.
edit: typos
Laptops with loud fans makes it really easy to find inefficient software.
Most people nowadays would just get a fanless computer and give up on fighting the bloat. But I appreciate your sacrifice, thank you.
I once noticed a memory leak in KDE when setting the background to a (very many image) slideshow instead of a single image. That was probably sometime around 2015.
Normally it's a bit tricky to calculate memory blame between X11, the WM, and the shell, since often their allocations are actually on behalf of an application, and killing the application will reclaim it.
KDE 6 isn't really that much heavier than a "full install" of xfce to be honest. I just prefer the simplicity (probably 1/10th the adjustability of KDE) of xfce for ease of maintenance, but KDE is fine too.
I mained dwm some years ago and I like it, but if I just need a quick environment I usually go back to Xfce. My family's computers (my dad's, my mom's, my sister's and my media PC, because I was lazy) are all Xfce. I will consider switching to Wayland when Xfce will support it.
I still remember seeing a video of Xfce (I think it was version 4.0.4) and being amazed how simple and reliable you could pick your own launchers and widgets on the bar. For me it was much simpler than Gnome at the time.
Modern KDE looks sometimes like Xfce for me (just my feeling, don't bite me), but somehow I always get a lot of strange bugs when I try to change things. Xfce had its own problems (I remember when Thunar was quite unstable), but it still is my go to desktop for simplicity and reliability.
>My family's computers (my dad's, my mom's, my sister's and my media PC, because I was lazy) are all Xfce.
same.
why?
they all grew up on Win9x and the visual metaphors used by XFCE just click, they could all care less what's behind the scenes as long as YouTube and Facebook work.
p.s. dwm is great. most suckless stuff is.
Yep tho kind of limited in configurability, it's really easy to modify the look of KDE for even casual users. I do wish the would at least ditch the default start menu and use whiskers or something similar though.
In all honesty, the screen sharing situation (and other stuff) in Wayland is currently so brittle that I consider XFCE (my daily driver) lagging behind in its adoption a feature.
I'm curious where you're having trouble. It seemed to me that this was quite bad at first (completely broken) and every application had to add explicit support, but I've had no problem with Firefox, Zoom and Slack recently. Just Microsoft Teams, but webcams don't even work in Teams.
Not the GP, but I have a workflow that depends on the functionality in x11vnc, which allows me to script sharing of individual windows, regions, or desktops. AFAIK, nothing comparable exists in Wayland, and there isn't even the possibility to bring something like that to non-wlroots environments. So I'm stuck on X11 for now.
There's really nothing that can be said then but I personally use `waypipe` like this `waypipe ssh user@127.0.0.1 sway` so that I can run remote sessions and it works amazingly. I sometimes use it to run GUI apps like thunar on my home laptop with tailscale when Im not at home. its basically `ssh -x`
https://manpages.opensuse.org/Tumbleweed/waypipe/waypipe.1.e...
Check out waypipe. It's not compatible with every piece of software, but when it works, it's like one of those magic "run x application over the network" legends except it actually works well.
I'm on wayland, have had no issue screen-sharing over teams. Webcam works for me, but only if it was plugged in before teams started.
wayland screen sharing has been solved like 4 years ago and would work already at the time with workarounds.
I remember when we got confined and were sent home, I could not screen share in MS teams on chrome out of the box but a few minutes of search gave me the chrome flags to enable and I just created a new local desktop file to start it up with the right flags
Since then the only things that have been missing is HDR support and color calibration. It sucks to not have them ready but they are only really used by a tiny fraction of the users so for 99.9% of the users wayland works fine.
I’ve got whole screen sharing working in sway, and I don’t dare touch it. Not sure if individual window / tab sharing is even a thing.
It is, but with a whole slew of tweaks to pull it off. It's a mess.
...I get it that Wayland is the future and all, that it has features we need as an OS to get to the next level, but I feel there should be no rush to do it, and I feel a strong push for Wayland in distros right now, even with all these little broken things all over the place.
Well I don't think the pace of Wayland introduction can be called "a rush" :)
I'm still on X11 myself for the time being as my OS doesn't support wayland properly yet (it runs but not with KDE)
Ditto. In terms of other stuff, I'm thinking about x2go, which I still use religiously? I'm curious, is Wayland anywhere on this?
The transition to Wayland also seems to correlate with the adoption of client side decorations (CSD). This "modern" approach destroys the traditional UX of XFCE as seen by recent changes in the settings manager. I fear for the future of XFCE. The advantage of XFCE for me has always been that it's a stable implementation of a traditional Win98/XP UX. I hope they don't adopt more Gnome3 patterns.
> The transition to Wayland also seems to correlate with the adoption of client side decorations (CSD)
Not really. The GNOME/GTK folks were already on the CSD bandwagon well before Wayland. Wayland compositors are free to draw their own decorations (and xfwm4 will indeed continue doing that once it's a Wayland compositor), and one of the actually neat things about Wayland is that there is a protocol that allows the compositor to tell applications not to draw their own decorations. (Whereas on X11 an app can tell the WM it will draw CSDs and the WM can't do a thing about it.)
Certainly GNOME has gone all the way to CSDs (IIRC if an app on GNOME doesn't draw CSDs, they get no decorations at all), but that has nothing to do with Wayland.
What has this to do with wayland or XFCE? The only major DE that is using client side decorations is gnome. This has been true since 2018:
> I heard that GNOME is currently trying to lobby for all applications implementing CSD. One of the arguments seems to be that CSD is a must on Wayland. That’s of course not the case. Nothing in Wayland enforces CSD. Wayland itself is as ignorant about this as X11. [...] In fact we created a protocol (supported by GTK) that allows to negotiate with the Wayland compositor whether to use CSD or SSD.
From https://blog.martin-graesslin.com/blog/2018/01/server-side-d...
Wayland is agnostic, and it's up to the compositor and application to decide what to do while operating under wayland.
CSD has nothing to do with UX. CSD just means the application draws the window controls and borders.
The application having the ability to change how the title bar works is an anti-feature for me.
On Wayland, the compositor can either let windows draw their own borders, or it can disallow it. On X11, if a window want's to draw its own decorations... it will just do it, and you won't be able to do anything about it.
Pretty much every single Wayland compositor follows this except for GNOME, who refuse to do so. It causes problems with windows like Davinci Resolve who don't draw their own decos. This leaves some windows without any controls at all lol. But this is a tangent and purely a GNOME problem.
> This leaves some windows without any controls at all lol.
Don't all WMs have shortcut keys anyway?
I mean yeah... But most people are going to have trouble with that, especially being unable to resize or move the window. I even saw a tech Youtuber try out gnome and they were baffled on why they couldn't move or resize davinci resolve whatsoever. Its almost shameful how negligent and hard-headed gnome is being.
They say they want software freedom, but in reality you have to use their software their way or you will suffer.
I gave up when even normal settings were hidden and I had to use Gnome Tweak Tool, but when I upgraded, I found all my changes had been wiped.
The application drawing window controls and borders becomes a UX problem when it doesn't do it right.
> CSD has nothing to do with UX.
Sure it does. Allowing windows to draw their own decorations, often in different ways and with different styles and themes, and not respecting the settings in xfwm4 as to what window-control buttons should be drawn (and where)... that's a huge UX issue.
Depends on your definition of CSD.
The process drawing it's own decorations? Technical decision that both Windows (tad complicated) and macOS already do, nothing to do with UX. There's nothing stopping this model from having the uniformity of SSD (as like on macOS), but it does on Linux due to multiple toolkits which don't all agree with eachother.
Combined headerbars? UX decision.
My accessibility needs are dead under Wayland. There needs to be shortcuts that can be emulated per application and global shortcuts.
Their security model won't allow this. I don't know how accessibility and automation tools are going to get by trying to support every wayland compositor.
It is probably much better if the compositor implements parts of the accessibility functionality. Then you can have other parts as libraries and in the toolkits, but the integration should probably be in the compositor.
I had a simillar take until I found keyd:
https://github.com/rvaiya/keyd
Global shortcuts are managed by the XDG Desktop Portal.
If accessibility is a serious concern, Linux is probably not the OS for you. It's so far behind in accessibility compared to Windows it's not even funny.
> If accessibility is a serious concern, Linux is probably not the OS for you. It's so far behind in accessibility compared to Windows it's not even funny.
Linux is fine with X11 per my requirements above. The biggest problem is the fragmentation caused by Wayland. I cant develop software that targets any reasonable range of desktop environments much less, the features needed. If Wayland was left and X11 was right the community needs something in the middle paradigm.
> I cant develop software that targets any reasonable range of desktop environments
I highly recommend Qt for this purpose.
Gtk has been a hot mess for the past few years with Gtk4 and libadwaita shenanigans. However, Qt has continued to evolve incrementally without dropping any backwards functionality.
It's incredibly full-featured and has amazing tooling. In addition to Qt Creator you also have the CLI tools for packaging and whatnot and KDE has Kirigami for components.
The nice thing is that Qt applications look pretty good in native environments like plasma, but they also look pretty good in Gtk environments, including older Gtk-3 desktops and newer Gtk-4 desktops. So you cover all your bases. You also get Mobile and Windows/Mac, if anybody cares.
The two real downsides are:
- GPL
- C++ only (ish). There's other bindings but they will not have the same level and quality of tooling. Gtk has less tooling overall, but MUCH more high-quality bindings.
> > I cant develop software that targets any reasonable range of desktop environments > > I highly recommend Qt for this purpose.
I really appreciate your comment. QT is a great framework doing a lot of heavy lifting. However, the accessibility software I help maintain needs the ability to know what windows are in the foreground, all applications running and emulating key/text/mouse input in any of those running applications. Wayland is antithetical to those features by design.
I can see your point with QT for standalone applications. However, with interacting with other applications I'm not sure QT can solve the issue.
If you target just one Wayland compositor, I think this is possible. I mean, each Wayland compositor is basically a display server. The trouble is the Wayland protocol itself doesn't necessarily provide all the tools for this (it does provide some, particularly in conjunction with systems like pipewire and portals).
But, for example, I think kwin can definitely do this.
This is because GTK is C, not C++. It's much easier to write bindings from C
Yes, I think the design of GTK and GObject worked out really well in the long run. On the surface it seems stupid to use C and then just "bolt on" a C++-like system on top. Why not just use C++? But clearly it's worked out.
Unfortunately, GTK has it's own set of drawbacks.
Do I understand correctly that they are building on top of wlroots? If so, I am pretty happy about this development as I might get to enjoy the solidity of XFCE apps in other WMs / compositors like Sway or River.
Yes, eventually anyway. I've done a little work on xfwm4, using wlroots, but it'll be another year or more before that's usable. For the desktop components that work on Wayland, we've mostly tested on wlroots-based compositors.
If I had to use a Desktop environment in Linux, it would be XFCE.
But, this Wayland thing is pushing me over to the BSDs. So far I have no abandoned Linux and I doubt Wayland will become mandatory until Firefox stops supporting X.
Don't look now: https://docs.freebsd.org/en/books/handbook/wayland/
Could be worse. The whole Gnome 3 and Wayland thing pushed me to macOS.
I sympathize, but unfortunately Tim Cook's stewardship of the Mac, from the embrace of soldered RAM and even storage throughout the Mac lineup, to the increasing annoyances of macOS either in the name of security (e.g., notarization, the popups that request permission to read user directories, which is really annoying when using LLVM's debugger in modern macOS) or to help advertise Apple's subscription services, have harmed the Mac experience. The issue of soldered RAM and storage in particular pushed me back to Windows after more than 15 years of using Macs (which is a shame since Apple's ARM hardware is awesome when it comes to performance and energy consumption); the last time I used Windows as my daily-driver OS was during the XP era. Unfortunately I don't like any of the mainstream desktop operating systems; Windows is annoying and loves to get in the way, macOS is becoming increasingly annoying with each release and is also tied to hardware with massive prices for RAM and storage upgrades with no DIY options like in the past, and the Linux desktop has a Sisyphean development cycle with so much churn. Instead of working on refining the user experience and polishing applications like LibreOffice to make them more competitive to Microsoft Office, it seems like every few years there are major infrastructure changes that require major code changes or even rewrites.
I'm quite disillusioned with desktop computing in general. Microsoft and Google are leading the way with "enshittification" (https://en.wikipedia.org/wiki/Enshittification), and the Mac experience in 2024 isn't like what it was in 2009 when the Mac was at its peak, in my opinion. The Linux desktop ecosystem suffers from a lack of funding relative to its commercial peers, combined with the "organization from chaos" that comes with bazaar-style development. The X11 to Wayland transition would be easier to pull off if the Linux desktop has Microsoft- or Apple-levels of funding and cooperation, but such funding doesn't exist, and part of the characteristic of the bazaar is that developers and teams of developers get to work on projects of their own interest and its the community that "votes" on projects by using them. The bazaar definitely exhibits freedom, but the major downside is that major changes in infrastructure (such as the X11 to Wayland transition) could take a lot of time and have major pain points (for example, the development of incompatible Wayland compositors and the fact that the X11 concept of a window manager doesn't exactly translate to Wayland's concepts).
I love free, open source software, but I'm drawn more to cathedral-style projects such as the BSDs and HaikuOS where there is a coherent vision for the system. Unfortunately the BSDs are dependent on the broader FOSS ecosystem for desktops, which is dominated by the GNOME/KDE/Wayland ecosystem (which also priorities Linux and treats other Unix-like operating systems as an afterthought, but that's another discussion). I'll need to re-investigate HaikuOS to see whether it can be used as a daily driver today; I've seen that a lot of attention has been given to HaikuOS in recent years.
I'm slowly working on a side project where I'm implementing my dream OS, which is influenced by various alternate paths of workstation and desktop computing explored in the 1980s and 1990s but were overtaken by Windows and Unix, namely Lisp machines, the Smalltalk environment, and 1990s Apple initiatives such as OpenDoc and the original plans for the Newton, which involved a Lisp OS (https://mikelevins.github.io/posts/2021-07-12-reimagining-ba...). But I have a busy day job and so this will take years of effort on the side.
I'll be horribly blunt here but I don't give a shit about any of that any more. It's irrelevant which is why open source is a complete joke for most people.
I, much as other users, just want my computer to solve the problems I ask of it, not turn into an activity in its own right or a maze of hosting problems and pain to get a simple task done. The job of the computer is to free us from slavery and everyone seems to have bloody well forgotten that.
macOS is my dream OS as it stands.
I can add a cucumber to my shopping list on my mac and when I get to the supermarket, it's on my phone.
I can share a calendar with my family and it just works.
I can listen to music on it and the same music is on my phone.
I can open and move pages around in a PDF without having to futz with anything or open a terminal.
I can literally write on my iPad with a pen and a PDF pops out full of my writing which I can send to people for review and feedback.
I can use Lightroom, Photoshop and Excel, which are far superior to any open source products and I do not mind paying for that privilege.
etc etc.
The hardware cost and upgrade inflexibility, neither of which I have found a problem either way, is insignificant anyway to the pain I would incur without it.
And that's what the community does't get: the last 20% is the hard bit and the above is the last 20% and it has never even once in the history of OSS got anywhere even remotely near that. No one cares about the political purity and ideology behind it.
> And that's what the community does't get: the last 20% is the hard bit
I think a lot of people in OSS do get that, but in many ways it's significantly harder to do that last 20% in an open source project than it is for a proprietary software company to do it.
But ultimately I don't care that much, honestly. If you don't get what you need out of open source, that's fine. You not using it isn't a problem for me or for my own open source work. As long as there are enough users to keep communities going and developers interested, that's all that matters.
> and the above is the last 20% and it has never even once in the history of OSS got anywhere even remotely near that.
That's provably false, and it's pretty annoying when people make uninformed absolutist statements like that.
> No one cares about the political purity and ideology behind it.
A lot of people do. No, not a majority, and maybe not even a largely sizeable minority, but it's very very far from "no one". If you don't, I guess that's fine. I hope your proprietary software continues to do what you need and want, and that your needs and interests continue to be aligned with what those software companies want to give you. That hasn't been true for me for a very long time now, but certainly people have different needs and different thresholds for it to matter. But if that changes, feel fortunate that all the work we've done will still be available for you to take advantage of.
> The job of the computer is to free us from slavery and everyone seems to have bloody well forgotten that.
Your use of the word "slavery" is a bit bizarre and overwrought, but it's funny you say something like this. Open source software is the only kind of software that's going to respect your freedom. Proprietary software never will. You're at the whims of its makers, and can never take control of your own computing destiny.
Like the sibling said, I guess that doesn't matter to you now. But maybe it will, someday, when the game changes yet again.
As I've said elsewhere, your data is the important bit. And at least in my case that is 100% portable. The winds of change are inevitable, but you can move house without too much hassle.
If think you can trust your data with a bunch of proprietary software companies, I'm not sure what to tell you.
> And at least in my case that is 100% portable
For now. Time and time again I see stories where people think they have redundant copies of everything important, and can reconstruct things when a company pulls out the rug out from under them, only to find they're missing something important.
Maybe you aren't. Great, good for you. You represent a teeny tiny percentage of the sum of proprietary software users. Again, I'm glad you believe you're safe and happy with the arrangement you have, but you're uncommon, and that's the entire point.
And for me, it's not just portability. Most companies are going to mine your data, use it to train AI, sell it to third parties, whatever they think will make them money. Or they just get hacked, like everyone does, eventually. You can protect yourself from that with open source. Not saying you don't still have to be vigilant, of course you do... but with proprietary software? Well, best of luck with that. Even with Apple, one of the few companies I think are reasonably trustworthy stewards of customer data, despite all the other issues I have with that company.
> No one cares about the political purity and ideology behind it.
No one cares, until they do. Everything you mentioned depends on a single company, which follows the commands from just a few people, who can one day decide to lead the operating system in a direction which makes it no longer your "dream OS", and intellectual property law makes it so that no other company can undo their changes.
> The job of the computer is to free us from slavery and everyone seems to have bloody well forgotten that.
We haven't, and that's why we prefer operating systems we fully control, instead of being bound by the whims of some foreign megacorporation, enforced by intellectual property law.
The other side of the fence is just as bad. Look at what Redhat and Freedesktop did to Linux.
While you might have the illusion of control, the exit plan to take control is to take on tens of millions of lines of source code and manage it yourself. That is not control.
Your data is the important bit. If you keep that portable and/or recreatable then it doesn't matter what tools you use.
> Look at what Redhat and Freedesktop did to Linux.
Well now you're just contradicting yourself. I'm not going to say RedHat and fd.o have been perfect all the time or have always made the best decisions or haven't pursued agendas that may have been at odds with others.
But c'mon. On one side you're saying that open source can never be good because no one maintains the cohesiveness and attention span to do "the last 20%", but then you rail on organizations that have tried to set and maintain standards, and reduce fragmentation on the desktop?
You can't have it both ways.
> I'm slowly working on a side project where I'm implementing my dream OS, which is influenced by various alternate paths of workstation and desktop computing explored in the 1980s and 1990s but were overtaken by Windows and Unix, namely Lisp machines, the Smalltalk environment, and 1990s Apple initiatives such as OpenDoc and the original plans for the Newton
Cool, I'll be following that. To me, one of the most interesting aspects of Genera is the command line interface, where commands know their arguments' names, types, defaults, and help strings. The commands can output nicely formatted data such as tables and, because they're just Lisp functions, they can share structured data directly rather than parsing and serializing ad-hoc text formats as is that's typical in the Unix world.
I've never used Genera; I was born in 1989 and thus was too young to experience Lisp machines. However, I've read a lot about them over the years, and I wish we had a modern analogue. I think the closest thing we have to Genera is the Squeak implementation of Smalltalk and its derivatives (Pharo and Cuis Smalltalk).
I wholeheartedly agree with you; a system based on structured data or objects would be much better than what we have today. What I've been dreaming about for years is a system where everything is an object, much like Smalltalk. Instead of typing "ls" in a terminal window, you would send a get_files request to the current directory, which would return a list of file names. It's possible to emulate the traditional Unix "ls" command, but interacting programmatically with the system enables so many other uses that's much easier than parsing text or calling POSIX system calls. Because the user is working with objects instead of text, not only would it be easier to write scripts compared to Unix shell scripts, it would also be easier for developers to implement GUI tools for the system. For example, imagine if word processing documents were represented by a class that handled all sorts of manipulations of the document, hiding the details of the underlying file format from the user and developer in most cases. It would be easier to come up with scripts for generating and modifying word processing documents without having to use the word processor's macros (if they're available in the first place) or without having to dive deep into the underlying file formats of the documents.
Oh, I'm also too young to have experienced the Lisp hardware. I've tried Genera in an emulator, though, and I've read some of the manuals because it's so interesting. There's a nice user interface programming manual preserved at [1]. Defining commands starts on PDF page 39.
[1] http://www.bitsavers.org/pdf/symbolics/software/genera_8/Pro...
I do not like the panels Apple uses, they have harsh backlight, due to 450 nm blue primary; nwer panels have 455-460nm blue light peak. My eyes are sensitive and I get tired on Apples.
Have you tried blue-block type lenses? Not those yellow ones from infomercial - it's a clear lens that cuts some wavelength for staring at bright monitor all day. Lots of cheap readers have those.
This was not my point. My point is that Apple uses old panels, which are pushier on eyes, instead of asking LG to put normal backlight. A supposedly premium manafacturer stuck with the panel from 2016; Apple stopped innovating, outside the M cpu series.
Btw the lenses still are yellowish, and interfere with white balance. Attempt to correct the white balance will destroy the effect of the lenses at 450 nm will be back.
> soldered RAM
Soldered RAM is faster than socketed RAM. What's faster still is Apple's current approach: putting the RAM inside the CPU chip package. This is one of the reasons why Apple Silicon chips are so screamingly fast and handily beat any x86 chip core-for-core.
> to the increasing annoyances of macOS either in the name of security (e.g., notarization, the popups that request permission to read user directories, which is really annoying when using LLVM's debugger in modern macOS)
These help prevent malicious software, such as downloaded apps or even web pages, from exfiltrating or tampering with user data. Relevant xkcd: https://xkcd.com/1200/
I'm on X11 still and completely ignored the train wreck, hoping it'll go away eventually.
X11 is basically already unsupported, pretty much all the developers moved to Wayland. Even the BSD and everybody else relaying on X11 has noticed that and they know they can't maintain X11 by themselves.
It will 100x not go away, there is no way. Literally all big stakeholder both open source (Gnome, KDE, Fedora and so on) and companies like Valve are all in on Wayland.
The reality is, X11 is a complete train wreck and always was. Its just that for a long time people invest in it and made it usable. The same is true for Wayland, things have been improving rapidly in Wayland and it many ways its already better.
The train wreck is being cleared and the trains are running again, literally nobody will go back to horses.
You say X11 is unsupported, and I note with glee that X11 hasn't _broken_ for me _since_ they stopped supporting it.
When software works it just works, and if you don't change it it continues to work. But if you had an issue, its unlikely that it would be fixed or that new features you want would arrive. Unsupported is fine for much software on the world.
Indeed. It is often a feature that core infrastructure doesn't gain new features.
Hardware is evolving, and while the core feature set works, it wont keep up. If you personally don't care about those things, that fine, and its a feature for you. But eternal stagnation isn't actually a realistic option.
yes, I wish more software would be "unsupported" like this!
I mean yeah. Wayland started with a feature aversion from its initial developer, then continued with feature avoidance by committee, the committee in turn required by the "everyone gets to implement the X server equivalent" design and some of the trouble coming particularly from "everyone" including Gnome. Some of these missing features are really not considered optional at this point by many desktop Linux users. But there aren't many left now. I'll probably make the switch in 0.5-3 years ;)
What is there in Wayland is really quite clean though, so it will... EVENTUALLY... be pretty good.
>X11 is basically already unsupported, pretty much all the developers moved to Wayland
In my view, Wayland is basically unsupported. GNOME and KDE support it. XFCE, Lxqt, Mate, and Cinnamon don't. It's been over 10 years. You could make a whole DE from scratch in this time.
> It's been over 10 years.
For most of those years, fundamental parts of the desktop experience were just not possible on Wayland. Even today, there are some current Xfce features that are just not implementable on Wayland.
> You could make a whole DE from scratch in this time.
Depends on your goals and team size. Part of me would love to rewrite Xfce in Rust (yes, I'm one of those people), using a different, lighter, actually-well supported toolkit (unlike GTK3), but with the team we have, that would be another two decades.
Consider that Xfce 4 is a little over 20 years old at this point. It's taken that long to build it to what it is. Starting over wouldn't take any less time to get there with the people and resources we have.
Have you looked at cosmic and iced? Maybe there is some bases of cooperation.
What features are missing from Xfce perspective?
> XFCE, Lxqt, Mate, and Cinnamon don't.
XFCE, Lxqt, Mate all have at least partial wayland support.
Cinnamon has experimental support.
That's not support. That's <qualifier> support. If that counts, then X11 having legacy support counts as well.
I don't agree, because experimental support means it's being worked on. Legacy support means the opposite - it's not being worked on and never will be worked on.
So for now it happens to work. But none of those DEs are contributing to the X.org implementation of X11, right? So it will slowly break down over time. That's the current status of X11, it's on hospice.
The current status of X11 is that it works.
The current status of Wayland is that it doesn't.
It's really that simple. I think this exchange can help you understand my position better: https://github.com/libsdl-org/SDL/pull/9345#issuecomment-201...
>Wayland has a myriad of unresolved problems regarding surface suspension blocking presentation and the FIFO (vsync) implementation being fundamentally broken leading to reduced GPU-bound performance.
To which someone replies:
>If we do this, we are basically accepting these issues are unfixable for the next ten years (SDL4). Having this as the default in SDL3 (which isn't being used yet!) is important for signaling to other stakeholders that we actually do need to get solutions in place for detected issues.
To which the committer says:
>SDL is not your tool for "signaling to stakeholders" about what is important. It's an actual library used by real developers and users!
Likewise. My PC is not a tool for signaling library/DE developers that there are more and more people using Wayland now and that they can finally stop maintaining X11 legacy code. I use my PC for real things. I couldn't care less if things are running on X11 or Wayland so long as they work.
If Wayland isn't working with my use case, I just won't use it. I won't switch to Wayland juts to beta test it, and then spend years begging people who know how to hack Linux to implement the features I need that I had on X11.
If Linux drops X11, and Wayland still doesn't work. I won't use Wayland. I'll use Windows.
> they can finally stop maintaining X11 legacy code
They already have, this is what I'm trying to tell you. All the X.org maintainers are on Wayland, and none of the DEs are contributed to X.org. It's on hospice. It happens to work just because, but that won't be the case forever.
It's fine if you currently like X and use X. I myself use X on Debian Stable.
But the reality is that it's dead technology, and eventually it won't work right.
Wayland is constantly being improved by many, many different parties. Valve, KDE, Gnome, freedesktop, Redhat... they're all constantly making new protocols and implementing them to solve problems.
They haven't stopped maintaining it because the programs still run on X11.
GIMP supports PNG. That doesn't mean GIMP contributes code to libpng, or libpng needs updates. It just needs to have code to make it work with PNG.
X11 will stop being supported when the developers literally remove the code to make their programs run on X11 from their projects. Until then, it's still supported.
Why would I need updates when things already work in the current version?
All X11 programs already run under Wayland. There will always be applications that are X11-only and they will always work. This is a fake problem, IMO.
> Why would I need updates when things already work in the current version?
I guess you don't, but then again, I don't see you running Windows 95. I think this is maybe something you like to believe, but in practice people do want new things and new software.
The only reason I don`t use Windows 7 is because Chrome, VS Code, Steam, and other electron apps stopped working on Windows 7. :-)
You maybe don't want to hear it, but the overwhelming majority of people are on Gnome or KDE. You might as well claim X11 is not supported because Hyperland, Cosmic and many others aren't supported on X11.
Unsupported in the context of software doesn't really mean what you took it as anyway.
> In my view, Wayland is basically unsupported.
"Unsupported" as in there are no maintainers, there's no roadmap, there's no future. X11 is a dead end. All the people who were working on it would rather work on Wayland instead. It's taking a while for the DEs, etc. to catch up, but even the slow pace of improvement Wayland is exhibiting is much better than "dead in the water" which is where X11 is.
It's not going to go away. Wayland is here to stay.
And so is X11. We are heading for a dual stack desktops. There are tons of expensive applications that target Linux _because_ of X11 and they will rather force users to switch OSes than get rewritten.
Pretty much any app that uses a toolkit won't need to be rewritten. Most of them just need small changes, or even no changes.
Those that are written to use Xlib directly, sure, they'll take some work (though rewritten? doubtful). If their maintainers don't want to do that, they'll still run under Xwayland.
These apps were written for X11 and NFS, so they can run on any node in a cluster. The were typically deployed on commercial UNIX systems and ported to Linux when Sun went belly up. In practice, I don't think porting them again will be needed. It is enough to qualify a different distribution if Redhat chose to exit the workstation market. There are already companies like NoMachine that keep X11 running.
It is not even about the access (people usually use nx or even vnc for that). It is about having a system that can work as a part of a cluster (network filesystem, X11), with all base GUI utilities (terminals, file managers, browsers, pdf viewers) supporting that.
> And so is X11. We are heading for a dual stack desktops.
IMO, we're heading for "X11 over Wayland" desktops, in which the only X11 implementation is Xwayland. These expensive applications which target X11 will keep working, through that compatibility layer.
If the apps target Linux/X11, wouldnt they also have to be rewritten to switch OS?
There are other Unix systems but by far the easiest solution is simply making sure all installed apps can still behave as X11 clients. Wayland or Xwayland is not really any issue. Few people use Xorg directly, as the new deployments usually require some form of a remote access.
I think it's trivial to run X11 applications under wayland, in a sandbox.
Sway works fine on FreeBSD.
I haven’t figured out the magic dbus (I guess) incantations required to make screen sharing work without systemd though. I guess it needs pipewire at least.
When I've had servers with a desktop env, it's always been FreeBSD and XFCE.
This "Wayland thing" is what the X devs would rather work on instead of X. It's the future. Best to rip the Band-Aid off now.
Perhaps if the X devs want people to migrate, they should consider giving Wayland all the features of X. Otherwise people are going to keep using the thing that works.
> all the features of X
I don't think Wayland should add a print server ;)
No, but there should exist a print server that Wayland applications can use. I don't object to refactors and replacements, but the underlying features should still exist.
There's 0 point in making a new software if you're just going to inherit all the fundamental design flaws that make the original software unmaintainable.
GPUs live in desktops and phones, but apps live in remote datacenters, so the future seems to be sending rendering commands to Javascript frontends on end users' browsers.
Pretty much, yeah. Hence why I say the replacement for X is Wayland, but the replacement for NeWS is the browser/Electron.
I used Windows 7 until Steam stopped working. The band-aid stays on.
I thought XFCE worked on *BSD as well?
Xfce is not perfect but pretty close (in *nix realm); it is a stable WIMP desktop environment. Everything where you expect it to be, no excessive feature creep, stable. It is like a tool, I love it. I wish it had a better brand. Too nerdy.
For anyone else not in the know
> In human–computer interaction, WIMP stands for "windows, icons, menus, pointer" —https://en.m.wikipedia.org/wiki/WIMP_(computing)
Hope XFCE still keeps xorg reliable
Most projects introduced bugs in xorg trying to support wayland, which is not reliable for many use cases, making everything unusable every other month
That's the plan, anyway. And I expect Wayland won't be usable for most as a daily driver until 4.22 or even 4.24, so obviously X11 will remain a high priority. And even after then, the X11 code paths will still work. Most of the difference is handled inside GDK anyway; the work we've done for Wayland support is mainly where GDK doesn't handle what we want to do, and where we've had to write libX11 code.
Most projects will eventually remove the X code path.
The long-term solution is to switch to Wayland.
I regret clicking this link without Adblock. Popover videos, mailing list signups, giant ads taking most of my verticle space, ugh.
Does anyone a cleaner article?
Ok, I’m super confused, how much work is involved in supporting Wayland vs Xserver?
[editing to add: my assumption that it isn’t is due to this post - xfce is what I used to use because it was lightweight, but if such a lightweight manager requires significant adoption effort it would imply that Wayland isn’t X API+new extension]
I had thought that Wayland was still functionally an implementation of the X server protocol, just reengineered to drop any pretense of network transparency, but given the amount of work involved that’s clearly not the case?
(It has been a long time since I did anything resembling X development, and so I’ve paid literally no attention to the details)
The core wayland does not even implement the idea of a window, to show how far the differences go.
Based on experiences of an acquaintance writing from scratch support for wayland for an UI framework that shares absolutely zero code with any of the major ones, the differences in even basic things like how to implement drop-down menu are staggering.
Funnily enough, said acquaintances needs would have been fully covered by slightly more capable XRender and a way to synchronize to VSync that does not depend on GLX.
The oversimplified answer is that a Wayland compositor must include all the functionality of an X window manager and the Xorg display server. There are libraries to help implement all that, but it's still a big lift.
Wayland is an entirely new display system that's completely incompatible with X11.
X11 apps can generally run under Wayland, but through XWayland, an X server that runs nested under a Wayland compositor.
Oh, ok, so X apps can run under Wayland, because Wayland DEs can run an X server (an existing X impl, or does Wayland provide an X translation layer/impl?)
But the DE/window manager is what is actually responsible for deciding when/where to provide that interface? For actual rendering+windows etc the DE is communicating entirely through Wayland specific APIs?
I guess the operation mode is similar to how you would run X apps on Mac and windows where you could launch an x server that would act as a layer between the Mac/windows windowing system?
Wayland is written by most of the same folks who maintain xorg, and XWayland is part of xorg.
It's generally more transparent than running a separate X server on Mac or Windows -- you don't normally need to care about whether an app is X11 or native Wayland.
Xwayland is basically Xorg with a Wayland DDX layer -- and it recently got a rootful mode. So that means it's possible to run a complete X desktop, including WM, on top of a Wayland compositor.
However, the drawback is that with the exception maybe of Weston, one does not simply start up a Wayland compositor. You need to have support for logind seats and that sort of thing. So unless you want to spend an afternoon assembling the components like a piece of flat-pack furniture from Target, you basically need systemd and a distro that bundled those components for you the right way.
> You need to have support for logind seats
Not really. It's perfectly fine to just advertise a single dummy wl_seat, and use whatever non-logind mechanism there is to tell the OS that you want to be in control of the input and display hardware.
There is no reverse equivalent of Wayland->X11 to my knowledge. Just the X11->Wayland via XWayland .
You can run any wlroots- (and possibly smithay-) based Wayland compositor nested inside an X11, but it won't be rootless; you'll get a separate window that all the Wayland apps will be drawn into.
There's really no reason why someone couldn't write a "WX11" or something of that nature, but I'm not sure it would be worth the effort.
Weston has (for the time being) an X11 backend.
A desktop environment in X11 is leaning on the X server. One on Wayland has to have libs for most of the things. There is no window manager. A compositor is like a server and a window manager in one. There are libs, but the process is much more involved and the API is much more tight so not everything is easy to get as in X.
Wayland is still broken in so many ways it's not funny.
Just recently I had to switch back to X11 because Wayland was just awful performance wise, like stuttering and constant fps drop. And this on a what 2023 high quality laptop with standard Ubuntu.
The way Wayland is done is a mistake.
This is great news!
I have switched my machines over to KDE for Wayland support and a bit better overall XP but still love XFCE for what it is.
Getting back to XFCE, and being able to leave gnome would be a godsend. I didn't dare to try KDE yet.
Ive got a hybrid XFCE/i3 setup as my main desktop.
The panel is killed on startup.
This could get me looking into sway/Wayland.
Should the project be renamed to WFCE?
No, not anymore:
Xfce is pronounced “ecks-eff-see-ee”. The name Xfce originally stood for “XForms Common Environment”, but since then Xfce has been rewritten twice and doesn't use the XForms toolkit anymore. The name survived, but it is no longer capitalized as “XFCE” and is no longer an abbreviation for anything (although suggestions have been made, such as “X Freakin' Cool Environment”).
Why should old WMs/DEs add support for Wayland, why not just write new for Wayland specifically?
Because for us (Xfce) that would probably take 15 years.
The only component that requires major changes is the WM. Otherwise, for the rest of them, it's more about fixing platform specific quirks.
(Ok, we did have to write an embedded compositor library for our panel to use, but... yeah.)
In this case, because XFCE is based on GTK, and the backend for GTK has gotten a lot of support for native Wayland recently. Starting from scratch would probably end up duplicating a lot more code than necessary, whereas supporting both in one codebase helps keep the featureset in lockstep for both x11 and Wayland users.
Why rewrite stable code? rewrite the layer that you need to rewrite. There are plenty of new DE for wayland if you're so inclined, however.