I think the problem with btrfs is that it entered the spotlight way to early. With Wayland there was time to work on a lot of the kinks before everyone started seriously switching.
On btrfs a bunch of people switched blindly and then lost data. This caused many to have a bad impression of btrfs. These days it is significantly better but because there was so much fear there is less attention paid to it and it is less widely used.
Both Fedora and openSUSE default to Btrfs. That’s all the praise it needs really.
With Bcachefs still being relatively immature and the situation surrounding (Open)ZFS unchanged, Btrfs is the only CoW-viable option we got. So people will definitely find it, if they need it. Which is where the actual issue is; why would someone for which ext4 has worked splendidly so far, even consider switching? It’s the age-old discussion in which peeps simply like to stick to what already works.
Tbh, if only Debian would default to Btrfs, we wouldn’t be having this conversation.
You are welcome to start a movement to get Debian to switch. You will be swimming up stream but you are welcome to try. Debian has been the same for decades and people like that.
You didn’t get my point. Btrfs is one OG distro removed from being THE standard. It’s doing a lot better than you’re making it out to be.
It’s not like Btrfs is dunking on all other file systems and Debian is being unreasonable by defaulting to ext4. Instead, Btrfs wins some of its battles and loses others. It’s pretty competent overall, but ext4 (and other competing file systems) have their respective merits.
Thankfully, we got competing standards that are well-tested. We should celebrate this diversity instead of advocating for monocultures.
It sounds like btrfs is solid most of the time and will explode for like 1 and a thousand cases.
A few years ago left my Fedora machine at home and left for a few days on a trip. When I got back the device was powered off and when I powered it on it said no boot device. When I booted off of a USB the drive showed as unknown with no formating to speak of.
I was able to recover it and the btrfs partition as apparently the GPT table had been overwritten. To this day I have no idea what went wrong. Btrfs in general is very solid in my experience and I use it for USB devices and my Fedora machines. I have never had a issue outside if that one time it died.
Btrfs is the filesystem that is cool but also potentially explosive. I think it has a huge amount of potential and I am very tempted to move my Proxmox machines over since it doesn’t have the same limitations of ZFS
@lancalot @possiblylinux127 I tried it once, it pissed itself and corrupted the entire file system to the point where I couldn’t recover, went back to ext4. Had similar experience with xfs.
what ZFS situation?
I’ll keep it brief. But it comes down to the fact that, out of the more popular distros, it’s only officially supported on Ubuntu.
https://openzfs.github.io/openzfs-docs/Getting Started/index.html
i’ve found to work without issue on Fedora, Arch and Ubuntu so maybe it’s supported very well
@lancalot @possiblylinux127 eh, also Garuda defaults to BTRFS, EOS does not default to BTRFS, but it has an option on their Calamares
I wanted to stick to (what I’d refer to as) OG distros; so independent distros that have kept their relevance over a long period of time.
But you’re correct, Garuda Linux and others default to Btrfs as well. At this point, I’d argue it’s the most sensible option if snapshot functionality is desired from Snapper/Timeshift.
@lancalot none of the “main” distros default to BTRFS, just “derivatives” default to BTRFS, Garuda is based on Arch, so it’s normal that it’s one of the rising new distros, Garuda rose because gaming on Linux received a huge boost from sources like Valve so I doubt that it (Garuda) will deviate from its path with time, plus, they provide multiple flavors for multiple purposes, gaming requires stability & sometimes a rollback mechanism, that’s where BTRFS shine, not so much stability BTW
none of the “main” distros default to BTRFS, just “derivatives” default to BTRFS
So you don’t regard Fedora (or openSUSE) as “main” distro?
@lancalot OpenSUSE is based on SUSE (created in 1994)
Fedora was developed as a continuation of RHEL
Maybe “main” is not well appropriate, I wanted to say “distros that have no precedence & not based on anything”, for example, 0.12 was a “main” distro, MCC Interim Linux was a “derivative” distroI suppose we differ in our definitions. Which is absolutely fine, to be honest*.
For completeness’ sake, IMO it’s basically the intersection of Major Distributions and Independent Distributions. Which happens to consist of Arch, Debian, Fedora, Gentoo, openSUSE and Slackware.
Out of these, Arch and Gentoo don’t have defaults, but their documentation uses ext4 most frequently for examples. For the remaining four, Fedora and openSUSE default to Btrfs. While Debian and Slackware default to ext4.
In all fairness, one might argue that Distrowatch’s list of major distros is arbitrary. Therefore, we could refine what’s found above by including actually data. For this, I’ll use Boiling Steam’s usage chart based on ProtonDB’s data. This ain’t perfect either, but it’s the best I can do. Here, we notice how both Gentoo and Slackware are not represented. Furthermore, NixOS poses as a candidate instead. For which, we find that (if anything) ext4 is the default. Regardless, it doesn’t actually impact the earlier outcome:
- Arch (and Gentoo) don’t have defaults
- Debian(, Slackware and NixOS) default to ext4
- Fedora and openSUSE default to Btrfs
Anyhow, what are the main distros according to you? Please offer an exhaustive list, please. Thanks in advance!
@lancalot the “main” that are alive today are (like on this graph) https://rreinold.github.io/explore-linux/ :Debian, Slackware, RHEL, Gentoo, Arch & android
These are only the alive ones, however, I couldn’t find any info about Nix OS so it remains on the maybe category cause I tried it and could not find any hint to the past
tbh the situation with Wayland was not too different, and wouldn’t have been better. Compared to Wayland, brtfs dodged a bullet. Overhyped, oversold, overcrowdsourced, literally years behind the system it was supposed to “replace” when it was thrown into production. To this day, wayland can’t even complete a full desktop session login on my machine.
So, if you ask me, btrfs should *definitively not * have been Wayland! Can you imagine if btrfs had launched on Fedora, and then you formatted your partition as btrfs to install Linux, but the installer could not install into it? “
brtfsreports a writer is not available”, says the installer. You go to the forums to ask what’s going on, why the brtfs does not work. The devs of brtfs respond with “oh it’s just a protocol; everyone who wants to write files into our new partition format have to implement a writer themselves”.Clearly you have had some bad experiences
Maybe you shouldn’t take your experience from 5 years ago and apply now. Wayland is solid and so is Btrfs. I know that because people use both.
I was mostly curious about btrfs with raid 1 on Proxmox but my doubts have been answered.
Except: I try Wayland every 6 months or so and still have problems with it.
Wayland’s problem isn’t Wayland; it’s all of the stuff that needs to work in Wayland that doesn’t. Using Wayland, to me, feels like using Windows, out a Mac: as long as you don’t stray out of the playground, it’s mostly fine (if a bit slow). As soon as you try to do any outside-the-box setup, like changing the status bar, things start getting all f’ed up. Like, last time I tried, I couldn’t get DPI font scaling to work - fonts would either be too small everywhere, or big in most apps but really tiny in the status bar. Whenever I encounter things like this, I search for solutions for, maybe an hour, see that other people have the same problem and there’s no fix yet, and bail back to X11, which Just Works.
Also, while I know some people have had bad experiences with btrfs, I’ve been using it for years. I originally switched because I had multiple separate cases of data loss using ext4, across different systems. It’s always baffled me that folks complain about btrfs, but ext4 was far less reliably. IME.
I have had the opposite experience from you with wayland and btrfs. Recent data loss with btrfs but perfect functionality with Wayland (on KDE and Arch Linux). Moving panels just works. Fractional scaling just works (though i do miss the old method where smaller screens just got supersampled instead of the way they do it now).
The key, probably, is that you’re using KDE - you’re playing “in the box”. I’m sure it works fine in that situation, or under Gnome; the desktops go to great lengths to make sure they work well under Wayland. Things get more dicey if you’re a WM user and are cobbling your environment out of multiple, independent programs.
I believe you about btrfs; enough people have complained about it that I’m convinced I’ve just been exceedingly lucky. I mean, by now I think it’s probably as stable as anything, but it seems like it used to have more issues.
@possiblylinux127 @lambalicious Wayland may be solid as a local display manager but it does not network.
It is a protocol not a display manager. The desktop runs everything and the apps connect to it.
Network was never part of the design and never will be
Removed by mod
I mean there is waypipe now …
Funny how you call people idiots yet it seems like you don’t know what you’re talking about.
I think most people aren’t living in the past. What is your use case exactly? What do you need a remote GUI for? RDP and other protocols exist and are much better especially in terms of performance.
@possiblylinux127 Again rdp, vnc, x2go, ONLY work for full desktops, they do not work for individual applications. If I’ve got a terminal session into a server and decide I want to fire up synaptic, X does that for me, Wayland doesn’t and the overhead of starting an entire desktop to run a single app for a few minutes does not make sense.
Hey holy shit! Someone else who knows this is in there! Two of us!
I got around this just using a local IPC on whichever box and a wlroots compositor (I think river right now? They’re all tiny). Wezterm is just locked in a remote multi plexed session I can fire scripts from the one with a keyboard. Startup isn’t really noticeable in wasted time, but it isn’t nothing. Consider this a promise to circle back if I do find something, it would have been a pain in the ass at another job.
Broadly I think it’s funny a lot of the same people who taught me the Unix philosophy don’t seem to understand the irony of refusing to move from the monolith. (Not that that’s you specifically, but Wayland actually gets it right for the most part).
The defense you hear from people like me is less fingers in ears that there are problems, and more the response to people who have tried nothing and are out of ideas on how to arrive at the same place differently. Im currently doing several things I was AGGRESSIVELY informed were impossible and wouldn’t ever be, but there’s so few people using some of the functionality in xorg I wonder if its back to hacking things together with pipes and scripts for the niches, remote display pretty much has been superseded by the common web server.
I started IT hearing stories of migrations back and forth from xorg/plan 9, and I did some of the troubleshooting in the early xorg era. you can trust me or not when I say I will choose the situation with xorg/Wayland now infinity times over that.
they do not work for individual applications
as someone else replied to you earlier, waypipe exists, and is packaged in distros, and does what you’re asking for.
There is also a newer thing called wprs, “Like xpra, but for Wayland, and written in Rust”: https://github.com/wayland-transpositor/wprs#comparison-to-waypipe which sounds promising
And there is were the community has kind diverged. Now days it is either headless servers or desktops.
Running individual apps is interesting but I am afraid that it is not super practical in 2024. However, there is this: https://github.com/udevbe/greenfield
Xorg is no longer being maintained for the most part and because the code base is so large there isn’t anyone who understands the codebase. I still use it for my semi virtual PC as Xorg allows for a lot more flexibility than Wayland plus Xfce4 isn’t completely ported yet. There will be a day when I move completely though. Probably when Xfce4 is Wayland native.
Tell me you don’t understand what “remote gui” means without telling me.
Maybe you shouldn’t take your experience from 5 years ago and apply now. Wayland is solid and so is Btrfs.
My 2 year old AMD-based laptop begs to differ. X11 is rock-solid, whereas Wayland locks up completely on a regular basis, without producing any useful logging. Every so often I try it to see if things have gotten better, but until today unfortunately not. Personally I prefer X11, I need to perform work on my Linux machine, not spend time debugging a faulty compositor, protocol or wherever the problem lies.
Wayland itself can’t crash, it’s just a set of protocol specs. The implementation you’re using (gnome/KDE/wlroots…) does. Obviously this doesn’t solve your problem as an end-user, just saying that this particular issue isn’t to blame on Wayland in itself.
Fine, in that case both Gnome and KDE handle the Wayland protocol in a crappy manner on my hardware. As the end-user I don’t care: I have no issues with KDE and Gnome on X11, when using the Wayland protocol they are unstable. For my use-case X11 is the better choice , as using the Wayland protocol comes with issues and does not provide any benefits over X11.
@Aganim @loutr This makes sense, these people that have some irrational emotion attachment to Wayland in spite of it’s lack of functionality, do not. Now, if they have a use case that makes sense to them, they’re playing a game that needs 200fps, then fine, but if the use case doesn’t fit then don’t use it.
This feels more like long time Linux guy digging in there heals because they like the old days
@possiblylinux127 @loutr I like to get work done, some tools are helpful to that end, Rust for example, superior to C in as much as it makes it much more difficult to make mistakes with memory allocation without resorting to the grossness of garbage collection, but when new things only detract from work flow, then yea I prefer the older things that work. When new things benefit it, Rust for example, or the latest kernels in terms of efficiency, then I use them. I don’t like change for changes sake, I like change when it improves things, in my use case, Wayland does not do that.
This sounds like a driver issue or something if all desktops are breaking for you. Have you tried reporting it anywhere?
The problem is, I wouldn’t know what to report and where. I’ve never been able to find any relevant logging, neither in
/var/lognor injournalctl. I doubt opening an issue with ‘desktop locks up randomly when using Wayland’ is really useful without any logging. And where would I do that? At the Wayland bug tracker? Gnome or KDE? Kernel, as it indeed might be a driver issue? And there is of course the time component: I use my laptop for work, so I simply cannnot spend hours on debugging this. That’s time I don’t have, I’m afraid.
Btrfs was solid for me some 11 years ago, Wayland still wasn’t solid as of yesterday.
Fam, my experience was one (1) (uno) year ago. And during those five years Wayland made zero progress by itself - it was everyone else who had to do the job of Wayland for free.
I think the main difference is that while a graphical session can work through some issurs, a file system is not allowed to fail under any circumstances. The bar is way higher and stability a lot more important.
Wayland didn’t work out networking, even to this day, which is why I’m still using Xorg.
waypipe?
X’s network transparency is overrated IMHO. Since ages most data on desktops is sent via shared memory to the X server (MIT-SHM extension) otherwise the performance would suck. This does not work over the network and so X over the network is actually quite slow. Waypipe works way better for me than SSH X forwarding.
@hummus273 It’s overrated because you don’t use it, I frequently do. If all you want to do is emulate Windows than Wayland is fine. If you need network functionality it is not.
You assume I’m not using it. On the contrary, I use it a lot at work. We have some old TK interfaces. They take ages to load over the network. The interfaces load much faster when using Xvnc running on the remote machine rather than X forwarding (but it is not as convenient).
@hummus273 Xvnc does not allow you to display individual applications only an entire desktop. I’m monitoring about 20 different computers doing different things and for me it is a significant advantage not to have to bring up a whole desktop but to be able to launch a single graphical application on my existing desktop.
I don’t really understand the degree of emotional attachment people have to one solution or another. For me it’s a simple application case, for me Wayland is not desirable, not only does it not network, but the embedded X-server as part of the kernel works very effectively by avoiding the kernel/userland switches an ordinary X server would require.
So for my use case, Wayland is NOT a replacement and so I have to object to people arguing that it is a full replacement for X, it is not.
@hummus273 Xvnc does not allow you to display individual applications only an entire desktop. I’m monitoring about 20 different computers doing different things and for me it is a significant advantage not to have to bring up a whole desktop but to be able to launch a single graphical application on my existing desktop.
Yes, that is what I meant with not as convenient.
I don’t really understand the degree of emotional attachment people have to one solution or another. For me it’s a simple application case, for me Wayland is not desirable, not only does it not network
Your use case is covered by waypipe (which in my tests is much more responsive than X11 forwarding).
the embedded X-server as part of the kernel works very effectively by avoiding the kernel/userland switches an ordinary X server would require.
I think you are confusing stuff here. Which kernel has an embedded X server?
So for my use case, Wayland is NOT a replacement and so I have to object to people arguing that it is a full replacement for X, it is not.
What part of your use case is not covered by waypipe?
@hummus273 Waypipe would involve a lot of userland / kernel exchanges avoided by using the kernel based mode setting Xserver. It happens to work well with my hardware. And I don’t see any noticeable latency issues and not all apps work with Wayland hence I have no motivation to change to Wayland and every motivation to avoid it. Sorry if that gets someone’s panties in a wad.
Waypipe has nothing to do with the kernel mode setting driver. The X server code does not run in the kernel. Wayland compositors use kernel modesetting for mode changes, so not sure what your point is? Not saying you need to switch to Wayland, just saying that it covers the use case you described as impossible with Wayland.
@hummus273 I have a 1gbit network connection at the co-lo, and 180mb/s cable and I don’t have any lag using X tunneled through ssh.
Not having any lag is physically impossible. You don’t notice it maybe. But if I open Firefox with X forwarding on the same network (1gbe) it is very noticeable for me.
@hummus273 Perhaps not because I’m not trying to game, and I can’t detect any changes faster than about 1/50th of a second anyway so fps faster than 50 is more or less moot for me.
Firefox is not a game?
Wayland as a protocol that apps use to talk to the desktop. It doesn’t use network at all really.
You need something like freeRDP for network access.
@possiblylinux127 It is touted as a replacement for X-windows but the PRIMARY ADVANTAGE of X-windows is that you can run a program on one machine and display it on anther making Wayland completely useless in a networked context.
It is not trying to be a one to one replacement. It is a totally different thing. You are wanting a motorcycle to replace your 2002 pickup truck.
Also X forwarding is broken for most stuff. It probably will work but it will run poorly and use lots of bandwidth. This is because there are layers and layers of work arounds to make modern hardware and software work on it. The X protocol was intended for mainframes in the 80’s. It should’ve died long ago.
@possiblylinux127 I agree with you, it’s not, but people here keep touting it as such and that is my issue. X-forwarding works great for me, I use it daily. I use it to access servers in my hosting service from my home office. I can fire up x2go if I need a remote console but most of the time a terminal and an app is quicker. I have no issues with lag, I’m sorry for those that do but it is just not a problem for me.
Do you use it over SSH?
@possiblylinux127 It strikes me as weird someone down votes a simple statement of fact. I guess they have a problem with reality.
With Wayland there was time to work on a lot of the kinks before everyone started seriously switching.
Not if you were using Ubuntu in 2017 when they switched to Weston as the default display server for 17.10 and lots of people suffered a great deal from how half-baked the project was at the time. For me personally, the 17.10 upgrade failed to start the display server and I ended up reinstalling completely, then in 18.04 they set the default back to XOrg and that upgrade also failed for me, resulting in another reinstall.
I have no doubt that this single decision was responsible for a large amount of the Wayland scepticism that followed.
People pretend Ubuntu is this great thing but in reality it hasn’t been great in 15 years.
@possiblylinux127 @drspod Expect a comment like this from Lemmy, bet you’re running Windows 11, I’ve got servers running Ubuntu 24.04, 22.04, 20.04, Debian Bookworm, Mint, MxLinux, Zorin, Fedora, Alma, Rocky, and Manjaro, the Ubuntu machines consistently give me less headaches even though I do have to purge them of snapd.
Out of all distros I’ve tried over the years, Ubuntu has always been the buggiest by far.
Not if you were using Ubuntu in 2017 when they switched to Weston as the default display server for 17.10
Do you have any source on Ubuntu using Weston as its default? As far as I know Ubuntu has always been GNOME, which doesn’t use Weston.
I stated the version number (17.10), the release notes are here: https://wiki.ubuntu.com/ArtfulAardvark/ReleaseNotes
Yes, the release notes you linked do not mention Weston at all.
CTRL+F, “wayland”, 1 of 3 matches
are you being deliberately obtuse or do you expect other people to do everything for you?
You said:
Not if you were using Ubuntu in 2017 when they switched to Weston as the default display server for 17.10 and lots of people suffered a great deal from how half-baked the project was at the time.
I said:
Yes, the release notes you linked do not mention Weston at all.
Unless you think Wayland is the same as Weston, I don’t see how you think I’m being “deliberately obtuse”.
@drspod @possiblylinux127 Since I am using Intel graphics and there is an Xorg X server baked into the Linux kernel for Intel graphics, I switched to it at that time and have been using it ever since.
Stop spreading disinformation (again). Wayland was a fucking mess and caused countless of issues, especially in a lot of “edge cases”. Meanwhile, dumbos were spreading lies about how it runs perfect and without issues while I kept switching back to X after merely minutes to hours whenever I tried to use Wayland again. It’s just bullshit that never was grounded in reality. Even now there’s games & applications who don’t run with Wayland, and likely never will since they have zero incentive to do so or aren’t even in active development anymore and that stupid X11 bridge still is required to run in the background for a lot of them.
I think you might be missing the part where wayland WAS running perfectly for them. It still does for me. I am actively and happily using Wayland and everything for me works. XWayland is a fantastic stopgap for now.
Wine is (slowly) getting a native Wayland port, which will translate to Proton eventually.
I had to switch to X after upgrading my system because it was all choppy on Wayland. I spent a few hours trying to figure out why and then I said fuck it, X still works. Wayland still is an unfinished mess.
Don’t use Wayland I guess
I just think people who have issues with Wayland and btrfs are hitting bugs that developers can’t trigger. It’s an unfortunate situation but these people should work closer with developers to get them more info.
The situation sucks because Wayland and btrfs offer so many features but there’s those few people who can’t seem to get them to work reliably.
