Is X.Org Server Abandonware? (phoronix.com) 123
Phoronix ran a story this morning with this provocative headline: "It's Time To Admit It: The X.Org Server Is Abandonware."
The last major release of the X.Org Server was in May 2018 but don't expect the long-awaited X.Org Server 1.21 to actually be released anytime soon. This should hardly be surprising but a prominent Intel open-source developer has conceded that the X.Org Server is pretty much "abandonware" with Wayland being the future. [Or, more specifically, that "The main worry I have is that xserver is abandonware without even regular releases from the main branch."]
This comes as X.Org Server development hits a nearly two decade low, the X.Org Server is well off its six month release regimen in not seeing a major release in over two years, and no one is stepping up to manage the 1.21 release. A year ago was a proposal to see new releases driven via continuous integration testing but even that didn't take flight and as we roll into 2021 there isn't any motivation for releasing new versions of the X.Org Server by those capable of doing so.
Red Hat folks have long stepped up to manage X.Org Server releases but with Fedora Workstation using Wayland by default and RHEL working that way, they haven't been eager to devote resources to new X.Org Server releases. Other major stakeholders also have resisted stepping up to ship 1.21 or commit any major resources to new xorg-server versions.
This comes as X.Org Server development hits a nearly two decade low, the X.Org Server is well off its six month release regimen in not seeing a major release in over two years, and no one is stepping up to manage the 1.21 release. A year ago was a proposal to see new releases driven via continuous integration testing but even that didn't take flight and as we roll into 2021 there isn't any motivation for releasing new versions of the X.Org Server by those capable of doing so.
Red Hat folks have long stepped up to manage X.Org Server releases but with Fedora Workstation using Wayland by default and RHEL working that way, they haven't been eager to devote resources to new X.Org Server releases. Other major stakeholders also have resisted stepping up to ship 1.21 or commit any major resources to new xorg-server versions.
Laziness (Score:5, Interesting)
Re: (Score:2, Insightful)
Re:Laziness Network transparency (Score:2, Interesting)
Re: (Score:2)
If you run XWayland on top of Wayland then you should have a server which supports the X protocol and is same order of efficiency, from the point of view of a networked client, as a normal Xserver. Is there any problem with that system that makes you want to run the X.Org server instead of XWayland?
Re: (Score:2)
I want to run Xorg because that's what all the VNC servers are based upon.
Re: (Score:2)
If you understand it better than me then why would you compile programs you want to use over the network against a library designed for use as a local accellerated graphics solution? What matters here is that there is no need to maintain the X.org server when the XWayland server does fine. Since the XWayland server doesn't need new device drivers it doesn't need to be updated in the same way as X.org. Added to the fact that you can actually do client isolation on Wayland by having separate X servers that
Re: Laziness Network transparency (Score:4, Insightful)
XWayland is part of the X.org so one could simply see Wayland as a back-end for X. The main problem is that there will be clients developed for Wayland that do not work (or not work well) for X. So we gradually loose compatibility with a decades old protocol. The other problem is that in my opinion Wayland is inferior because it is much less flexible and this will bite us later. X is an extremely flexible protocol for remote graphics buffer management and communication between clients. The people who mention some ancient feature of X and claim that X is bad because supported printing or server-side fonts got it all wrong. That X was able to evolve for such a long time shows that it is based on an excellent design! You can do client separation on X as well, either by running nested X servers or - which would actually be much better - by using untrusted X clients. If Redhat / Canonical would spend their money wisely: We could have: excellent and efficient remoting (by using xcb with asynchronous use of the X protocol), the possibility to connect and disconnect individual windows from the xserver (just requires toolkit support), the ability to move windows from one display to another (even on different computers, also just requires toolkit support), the possibility to use most clients as isolated untrusted clients (which would be real advance in security) while still having working cut&paste etc. In other words, a windowing system ideal for networked world..
Re: Laziness Network transparency (Score:2)
Re: (Score:3)
Is Wayland network transparent, like X? My understanding is that it is not. If that is true, then I see Wayland as being more like MS Windows, and without network transparency, it is effectively useless to me, as I depend on the ability to run a client on a remote machine, and have the winnows it generates pop up on the X server on my workstation. Or have a machine running industrial control software, and pushing it's displays out to operator workstations running X servers.
Somehow, it seems like every single new project advocated by someone declaring the death and obsoleteness of X11... Completely and utterly forgets about network transparency.
Of course it doesn't help that the distinction between "remote while-friggin-desktop" and "remote single-application" is completely lost on people who have only experienced what Microsoft (and VNC) gave them.
Re: (Score:2)
Why do we need an entire UI stack based around a feature a handful of its users are ever going to use? Thats why X replacements skip network transparency - its not important to most of the user base.
I havent used VNC in a decade, I last used RDP four or five years ago, I havent used X11 in 20 years and I dont miss it. And Im firmly in the “power user” bracket - I just dont need network transparency.
It might have been useful 30 years ago, but today its a niche requirement.
If you need it, use so
Re: (Score:2)
I use remote network sessions in my environment in order to access secured network segments like the DMZ. I can use RDP or VNC to get in to specific devices, but everything else is firewalled off. So I set up a workstation inside the zone, security hardened of course, whose sole purpose is to be able to access the rest of the machines inside the zone. RDP and VNC are still useful for that at least.
Re: (Score:3)
Re:Laziness (Score:5, Insightful)
Please volunteer to help work on this. The money you are paying (zero) is not enough to support the needed development staff.
Re: Laziness (Score:3, Interesting)
We aren't anywhere near the biggest paying customers for this stuff. "There's no money to maintain it" is a load of shit and has been for decades with regard to X11 in particular and OSS in general.
Re: (Score:2)
Maybe a crapton of money isn't enough to justify working on this.
After all your employer isn't willing to pay someone to do it. It's open source, they could get someone to add the features they want, but apparently it's not worth the cost.
Re: (Score:2)
Fork out and do it yourself then.
Re: (Score:2)
Comment removed (Score:4, Insightful)
Re: (Score:2)
Input and capture are still too restricted last I heard for things like Barrier and screen sharing/remote control, I haven't heard a solution for that.
On the other hand, Xorg has security problems such that anything can screen scrape and keylog.
Interestingly enough, Xpra largely hooks applications through compositing and window management hooks, rather than the remote X calls (it lets a dummy X server serve the X call requirements but does not use them). It provides a better experience than X forwarding wit
Re: (Score:2)
However [Xpra] can't support Wayland yet because of the aforementioned security restrictions on input.
Why would that be an issue? A Wayland version of Xpra wouldn't be trying to intercept output intended for another compositor, or inject input into another compositor, the way it does with X11; it would be the compositor. Apps would connect directly to Xpra to present their frames, and receive their input directly from Xpra. There are no security issues in the way since Xpra would have complete control over presentation and input. There are tutorials for implementing a Wayland compositor with the wlroots lib
Re: (Score:2)
I may not have it fully understood, just looking at:
https://xpra.org/trac/ticket/3... [xpra.org]
And assuming the lack of capability would be security related somehow, rather than just 'missing' (since it's gone a long time to still be missing).
Maybe they just got it right the first time? (Score:4, Interesting)
Perhaps it's just all perfect?
Re:Maybe they just got it right the first time? (Score:5, Insightful)
Re: (Score:3, Interesting)
>"X11 is probably not perfect, but certainly way more function-complete and mature than that infantile experimental toy called "Wayland"."
Which is something I have to explain to people wondering why development on Xorg has stalled. It is so mature and so robust, it doesn't really NEED much development. It could use some upgrades, but that mindshare switched to Wayland, which I think is very unfortunate (because I don't like that model at all). I was hoping one day to even see X12. I don't think that
Re: (Score:2)
The decade of work wasted on Wayland could have definitely been used to improve X. Better use of the asynchronous nature of the protocol would make it much faster over the network. Connect- and reconnect support. The possibility to move windows from one display to another etc. All this is easily possible with the existing X protocol but not (well) supported by most clients. Client separation (make more clients work as untrusted, which would also bee easily possible).
Instead we get a stupid replacement which
Re: (Score:2)
What specifically would you like to see upgraded in X.org?
Re: (Score:2)
>"What specifically would you like to see upgraded in X.org?"
That is a good question. One I can't really answer, because it does everything I need. But apparently it doesn't do what everyone else needs, hence Wayland.
Re: Maybe they just got it right the first time? (Score:2)
Either that (Score:5, Interesting)
Re: (Score:3)
Or it is a stable and mature implementation
That was my immediate thought when I read the headline. Think of the C programming language. It just works, for close to the hardware embedded applications, and for Linux kernel code, and so on. I do not want the rules changing.
Ya, but ... (Score:2)
The problem with anything involving Weyland [fandom.com] is that it's more interested in Xenomorphs [wikipedia.org] than ...
What? "Wayland?"
Never mind.
Can we just get rid of X (Score:1)
I hated it since it required half a PC's resources 20 years ago and can't do things a modern platform can do without hacks. Even my phone can do don't rendering without requiring a special font server. My last bit maybe outdated but was an issue 11 years ago when I last ran a gui on Linux.
Re:Can we just get rid of X (Score:5, Informative)
Your phone required a font server? I haven't used a font server with X for at least 15 years. And I have no idea what you're referring to about X consuming half your resources. I've actually used X11 on embedded devices and pre Android phones and it was pretty lightweight. Part of the problem with ongoing X11 maintenance is we only use a small part of the X server these days. No one is abandoning the remote X protocol so you can remote any modern gnome or kde app today with Xwayland running as a lightweight X server on top of Wayland. So maybe that's the best of both worlds.
Re:Can we just get rid of X (Score:5, Informative)
X11 is so lightweight that even the bastardized version we used to use in Cygwin to get X servers running in Windows XP didn't consume much overhead, and that was with the X client running on the same Windows XP box. When the X client was running on a *nix box, it was pretty damned tolerable. Perhaps the poster is thinking of desktop environments like KDE which definitely could run pretty darned slow and chewed up a helluva lot of resources, but that has little to do with X11. A lightweight desktop like XFCE is incredible in its responsiveness and speed, and frankly, just a cleaner experience than more "feature-rich" desktops like Gnome, Windows, KDR and the like.
Re: (Score:3)
Last time I tried using Cygwin was on one of the 486 clones (cyrix I think). Cygwin was a pityful experience in its own right and that was before you added X. I am not sure what you are talking about but X on cygwin used to be a nightmare.
You might be too young to remember this:
https://medium.com/@donhopkins... [medium.com]
It was written in 1994 I think and reflects the inefficiencies of X well.
However, ironically this type of inefficiency is no longer an issue. Hardware did become fast enough to run X and the XFree86 p
Re: (Score:2)
I never said getting X running under Cygwin was easy, although by the mid-00s there were compiled X client and server binaries. Cygwin was pretty awful, but unless I wanted to pay big bucks for a closed source X server, that's what I had to do. Honestly, as I recall, it took me an afternoon or so of banging around in configs to get everything going, and ran the X client on the machine just to avoid any potential networking issues. Once I had things running I just started up the X client on my Slackware box
Re: (Score:2)
Re: (Score:2)
Anyone using "face it" is making an emotional argument, not a factual argument.
VNC does not replace application level X11 network transparency, which is one of its most useful features. Nothing in Wayland does, or allows it to be replaced, and that appears to be by design.
Face it, Wayland is broken by design. ;)
Re: (Score:2)
I'm ok with a rethinking. I'm not ok with throwing away a feature that I and countless others need, whether you think it is X's "worst feature" or not, and whether it truly belongs inside the server (which I tend to agree it probably doesn't), or not.
Network transparency is absolutely mandatory for my work. Desktop sharing via VNC or RDP doesn't cut it. I'm fine if it has to happen in a separate layer and not in Wayland itself, but it has to be there, somewhere, and without the need for recompiling clien
Re: (Score:2)
Fully agree ...
I was on KDE for many years (Kubuntu).
There was one release that was botched, and the KDE developers blamed the distro for it.
Made me nearly abandon KDE.
Then the next release fixed that.
However, a few releases after that, when they started to remove configurability (one of the main differences between the
Re: (Score:2)
Re: (Score:2)
I was happily using Wayland on Debian stable for almost a year, then COVID required me to teach classes in Zoom everyday, which is not supported in Wayland. I switched to X.org and everything just works. I don't have anything personal against Wayland, but if I abandoned X.org I would have to abandon Linux altogether.
The answer is mostly yes (Score:4, Informative)
Large parts of the X.org server have been abandoned for many years. The font server, the built-in widget set, to name a couple of parts. I'm sure various input methods were tried over the years and abandoned. The modern X server is certainly a hack at this point. It works remarkably well, but it is still a hack and maintenance is difficult. The fact is very few people know how to fix bugs in the server. Fewer still understand the architecture. Several years ago there was good talk about this: https://www.youtube.com/watch?... [youtube.com].
Abandoned by developers or not, the X server is still supported by major distributions, and as far as I know is the only thing supported by nVidia's drivers.
Wayland as currently implemented is missing some key features like application-level remoting and screen capturing (by design it seems. Also Wine cannot on Wayland outside of a desktop window currently. But these issues are not insurmountable. For whatever reason, Gnome on Wayland could crash and bring the entire desktop and all your apps down, at least until recently. This is not an experience that is familiar to X11 users!
I use X11 forwarding over SSH all the time. Remoting is a key feature of X11 that I depend on. Wayland has no built-in remoting, which is concerning. But as long as the X11 protocol is supported by applications and GUI toolkits, it will always work on Wayland with the Xwayland light-weight X server. So with this, yes, even forwarding over SSH works with wayland.
Re: (Score:2)
Wayland has no built-in remoting, which is concerning.
It's not concerning because it's not a feature that is intended on operating at it's level, for good reason, it was designed singularly with performance in mind and adding this to the compositor puts performance at risk for what is ultimately very few use cases which could be met differently. There has been demo code showing that remoting was possible both VNC and RDS style some 8 years ago shimming on top of Wayland. They released it at the time to shut up the critics who claimed that they fundamentally br
Re:The answer is mostly yes (Score:4, Interesting)
And that answer has never been satisfying because the use case for X11 remoting is per app. Not a desktop. The idea that Windows-style remote desktop is somehow a replacement for traditional X11 forwarding is mystifying to me. That's not remoting. That's simply screen sharing. Sure that's useful in and of itself, but it's no replacement for X11 forwarding, which like said, isn't going away thanks to Xwayland.
Re:The answer is mostly yes (Score:4, Informative)
>"And that answer has never been satisfying because the use case for X11 remoting is per app. Not a desktop. The idea that Windows-style remote desktop is somehow a replacement for traditional X11 forwarding is mystifying to me. That's not remoting. That's simply screen sharing. "
Indeed. X11 is extremely useful in such ways.
>"Sure that's useful in and of itself, but it's no replacement for X11 forwarding, which like said, isn't going away thanks to Xwayland."
That isn't the way to look at it. It is only useful as long as the apps are using X11. If you code the app to NOT use X11, then that ability instantly disappears and Xwayland is of no value at all. That is the danger. Wayland is not a threat to X11 as long as the apps are X11. The moment something major, like Firefox, decides to switch to Wayland, everyone that uses X11 for remoting or for X11-based thin clients, is screwed.
Re: (Score:2)
The moment something major, like Firefox, decides to switch to Wayland, everyone that uses X11 for remoting or for X11-based thin clients, is screwed.
So Devuan was founded on the principle to support people's specific use case who wanted Debian without systemd. Where's someone to step up and write a RDP client/server that supports per app remoting (which is already a part of the RDP protocol)?
This "this program screwed all these other users on account of it existing" argument is precisely the bullshit that open source software is designed to avoid. The question is, how many of you are there, and how many resources are you willing to dedicate to getting
Re: The answer is mostly yes (Score:2)
Shush! Please don't give them any ideas! The way they're running around trying all kinds of new features to get their user base to abandon them is already astounding...
Re: (Score:2)
Re: (Score:2)
And that answer has never been satisfying because the use case for X11 remoting is per app.
I never claimed it was satisfying. I said it's a design goal. Wayland is the Ferrari of the family car industry. They have a goal to do one specific thing (and be good at that ... where have I heard that before). Sure some customers will complain about the lack of 6 seats, and a tow hitch but that's not the design goal.
Honestly the "that's not remoting" thing is a lack of interest in your use case. RDP supports per application remoting. It has for many years. The reality is so few people are interested in i
Re: (Score:2)
Sure, but when they stop offering the Honda Odyssey/X11, then what? It's puzzling why developers seem more focused on mythical new users and their user cases in other operating systems than on the needs of current users.
Of existing users of Linux, certainly long-time users of the kind who are or were system administrator types back in the day, forwarding over ssh is incredibly common.
And actually there is at least one brave soul who's stepped up and started work on waypipe. Hopefully that gets more love b
Re: (Score:2)
Screenshotting is now the responsibility of the compositor, so Weston allows you to screenshot if you wish but other implementations may do different things. It may have been possible to screenshot in X11, but that was also a well known security problem.
Remoting is also out of scope of Wayland protocol and is left to the compositor. e.g. Weston has a desktop RDP implementation but RDP-RA
Can wait for the year of the linux desktop (Score:2)
Re: (Score:2)
Still works (Score:5, Informative)
I'm not seeing any issue with it, so what does Wayland give me that X doesn't? Aside from a whole load of new bugs, of course.
Comment removed (Score:5, Interesting)
Re: (Score:2)
Did X got removed from distributions when they started offering Wayland? Is systemd only one of options offered at installation time? Does Wayland replace 10 times more then just X?
Bit misleading (Score:4, Informative)
Re: (Score:2)
Indeed. Software seems to now be interpreted as abandoned when it isn't constantly churning. As one asymptomatically approaches zero bugs and no further feature requirements for what ever reason it becomes a violation of natural software development laws and must be outcast as dead.
Yes, there's a lot of possible things to clean up with X.org but at the same time it's damned nice to not have the constant churn of updates.
Re: (Score:2)
X dates back to 1996, so is probably less than 1% of its life. It is probably stable code.
In Wayland's lifetime, that is the equivalent of about 8 weeks.
Nothing to see here. Move along please.
Berlin display server? (Score:2)
A long time ago there was work in a display server I think its name was Berlin something (consortium/collective)? It had everything X11 had. I been looking for information on it but cant seem to find anything about it now. I think I remember the name correctly. Seeing this post I'm now curios about it. I remember it worked on primitives instead of pixels. Is there anyone that remembers it?
Re: (Score:3)
Re: (Score:2)
It is fun to see how things still look pretty much (Score:3)
Re: It is fun to see how things still look pretty (Score:2)
Wrong Metric (Score:4, Interesting)
Looking at release frequency to determine if something is abandoned is the wrong metric. Yes, frequent releases would show that the project is active and alive, but what is more important is response to bug reports. Look at the bug tracker. Are there new issues since the last release that aren't being addressed? Is the time from but report to analysis going from days to months? Are there significant incompatibilities with new versions of packages that it uses, making it difficult to keep the system running (like Python 2 dependencies or failing to compile with gcc10)?
If you just go with lack of new releases, /bin/true is abandonware, but I'm pretty sure there are no bugs filed against it.
Re: (Score:2)
Well, the debate over "uncertain" vs. "maybe" is holding it up.
Re: (Score:2)
Well, the debate over "uncertain" vs. "maybe" is holding it up.
Are you sure?
Waypipe (Score:2)
https://gitlab.freedesktop.org... [freedesktop.org]
Re: (Score:2)
What would the update include? (Score:4, Insightful)
Most industries have the concept of a product being complete. Why the urge to update software for no reason?
If enough people wanted changes, there would be changes in progress.
Re: (Score:2)
Software is rarely complete. It generally must evolve and adapt to stay alive. When something ceases to do so, and especially in the face of numerous bug reports and/or feature requests, it's fair to ask whether it is still truly alive.
I think in this case, reports of the death of X are premature, but, possibly, not by much. When and if Wayland or something else evolves into a complete and adequate functional replacement for the parts of X that are in use today (including IMO network transparency, at lea
Re: (Score:2)
An important part of software development is annoying the end users.
It is a kind of "click-bait" used to ensure large amounts of Tweets(cf Trump).
And all I want is... (Score:3)
An X Server for iOS.
Re: (Score:2)
I think there are some in the App Store, but they aren't free.
Common sense? (Score:4, Insightful)
Common sense seems to have widely left the world of software. Common sense would say that a widely-used thing that nobody sees the need to sand here or polish there might just be mature. In software, though, things are prone to be called 'abandoned' or even 'dead' as soon as they stop to be constantly modified in one way or another, whether for better or for worse.
I see a reason for that in commercial environments where money needs to be generated by selling new versions every now and then, and where developers might not want to lose their jobs, but I really cannot see how it would fit free and open source software.
Xorg size (Score:2)
Xorg uses 900 MB RAM in my Ubuntu laptop; Can we have a stripped down version?
Well, at least X usually works for me, ... (Score:2)
Xorg current alternatives (Score:3)
Until a real alternative appears - Wayland does not cut it - Xorg is where it's at for me, in combination with X2go (which is based on nx) for a working remote service. Even though X carries around a big load of cruft I can just ignore that load and use the parts I need.
A real alternative for Xorg has to offer at least the same advantages - including network transparency which I use every day - and has to offer a complete backward compatibility layer for the foreseeable future given the plethora of existing
Re: (Score:2)
I'd given up on NX and its successors; the original NX was *amazing* and they fucked it all up afterward. Nowadays I use xrdp and any RDP client; it's pretty snappy, multiplatform and open protocol. I still hate that I can't export a single window like I could with the original NX, but it does everything else pretty nicely.
Re: And... (Score:5, Informative)
If you want to compare X11 and Wayland, you need to include all the many libraries required to add up to X11's feature set before you can start.
Re: And... (Score:4, Interesting)
Thank you! So many people seem to have know idea what these two products do. X11 is still the core of a lot of stuff simply because nothing else does what the X11 protocols do.
Re: And... (Score:4, Insightful)
If you want to compare X11 and Wayland, you need to include all the many libraries required to add up to X11's feature set before you can start.
That's not a very user focused approach. Was the car useless because it didn't run on hay and couldn't drink water? When comparing replacing one software with another looking comparing features is irrelevant. The only thing that matters is how each program meets the *user* requirements.
X11 has some good stuff that Wayland is missing. But it also has some features that would leave people dumbfounded. E.g. the ability to send the output frame by frame to a printer. A good modular feature, and one I would argue should never ever be a design feature of any GUI ever again, not even by accident. X11 is like running the drain from your kitchen sink into another kitchen sink. It is feature rich to the point of absurdity and and actively hampers its own progress as a result.
I will always have fond memories of it's network client server operation though. Modern systems don't quite compare, ... but then X11 no longer compares to what X11 was either.
Re: (Score:3)
That X has some features which today do not make sense does not matter at all. It is an extremely modular design which is excellent. Throwing it away and replacing it by something much less flexible is really... unwise.
Re: (Score:2)
Re: (Score:2)
Maybe this is sometimes true. But for X, I can not see the problem that justifies replacing it with a less flexible and powerful solution.
Re: (Score:2)
That X has some features which today do not make sense does not matter at all. It is an extremely modular design which is excellent. Throwing it away and replacing it by something much less flexible is really... unwise.
You're assuming that flexibility is free. It's not. To reuse a car analogy from another thread: The Honda Odyssey is incredibly flexible. It can carry 8 people, or fold down 6 of the seats for phenomenal cargo space. So why doesn't the Ferrari Portofino offer that flexibility?
Flexibility is not some design element that people can implement by choice. It is a feature in itself that needs to be traded off against other features (such as simplicity and performance). The lack of simplicity and performance due t
Re: (Score:2)
The only overhead from using an extension X is the query request during setup. Thus, all this statements such as "incredibly patchwork of stung together band-aids" that make X a problem (I do not have any problem, thank you) is just FUD. So it would be nice if you could make precise technical arguments and not vague general statements.
Re: (Score:2)
Err that is blatantly false. There have been plenty of analysis and benchmarks to show that a) the drawing overhead is significant, and b) the software development complexity is insane to say nothing of the incredible amount of effort required by maintainers to actually put a system together that will actually work on your hardware without shitting itself due to a multitude of different programs working with other programs the selection of which is depend on the specific program you chose to use.
e.g. Someon
Re: And... (Score:5, Interesting)
It's like sysInit. Everyone hated it and when SystemD came out for it's replacement everyone afraid of change or had years of work on cryptic rc files linked to shell scripts with thousands of lines of if/fi logic (a no no to put logic in a config file) they all go ballastic!
I suspect you haven't tried to adjust how systemd interacts with some novel configuration requirement. It's got cryptic nailed down good and proper.
Systemd is fine if it does what you need out of the box, so it's good for the wife's laptop which is never going to be used for anything at all off the beaten track. Otherwise it's no better than rc and config files and has a huge design error as its very foundation.
Re: (Score:2)
Complete hogwash.
I sometimes get servers with nonstandard setups and they are easier, not harder with Systemd. The dependencies are more reliable and if there is extended logic needed, you can still call a script for that part, only now it doesn't need to worry about dependencies, only it's own stuff. I have some mount points that are working with far fewer hacks than when everything was RC scripts.(Clustered FS mounts, FCAL)
Re: (Score:2)
It might just be inexperience, but I find myself fighting with systemd constantly. I'm fortunate not to have to in Gentoo Linux since it's still based on OpenRC. OpenRC does not manage my hostname, date/time, home directory, network, cron jobs, or anything else; it defers to other software do to these things, and just does its job, which is to start stuff. At work, however, systemd tries to manage all of the above, and requires that I jump through hoops if I disagree with the way it chooses to do those t
Re: (Score:2)
It might just be inexperience, but I find myself fighting with systemd constantly. I'm fortunate not to have to in Gentoo Linux since it's still based on OpenRC. OpenRC does not manage my hostname, date/time, home directory, network, cron jobs, or anything else; it defers to other software do to these things, and just does its job, which is to start stuff. At work, however, systemd tries to manage all of the above, and requires that I jump through hoops if I disagree with the way it chooses to do those things. It's easy, for me at least, to render a system unbootable, or at least un-networkable (which since I manage many remote systems might as well be the same thing).
I truly believe in the traditional UNIX philosophy of tools that do one thing well and can interact with other tools via pipes and such. If I wanted my work systems to be like Windows, with a single central software layer managing everything, I'd just run Windows. I really don't. Systemd in my view makes Linux more and more like Windows: perhaps better for the "average" desktop user, but absolutely hostile toward anyone and everyone else.
This is my experience too - I'm on Gentoo at home and Ubuntu at work and Gentoo is a lot cleaner IME. There may well be situations where systemd's complexity pays off but it remains a big fat single point of failure.
Re: (Score:3)
Virtual Machines, and Amazon E3 clouds. It also requires planning the images for things like a hacking attack or server going down. When an event happens how can you program SysInit to be aware?
Most people just create a new VM when there is a problem, sometimes with a rollback. Messing with init systems isn't worth the extra effort for a very minimal improvement in uptime.
Re: (Score:3)
With Systemd, my systems quite often won't work at all, let alone seamlessly, and it is not like debugging is supported by systemd.
Yes, I am migrating my servers to BSD and my workstations to Devuan.
I plan to make a Youtube video of pushing Ubuntu off a cliff.
Re: (Score:2)
Wayland support network transparency? (Score:2)
Re: (Score:3)
^--- this definitely. So long as ssh -X is supported, I'll be content; without that, I'm not going anywhere.
Re: (Score:3)
Okay, I think some have misread my critique...
It wasn't about what was good or bad or in need with X.org. That's a discussion that I'm sure we could all participate in and only good things would stem from it.
My critique was on the article itself.
All he did was try to claim that it is dead, that nothing was going on with it, so just move on.
There was nothing helpful in there. No pleas for reviving the project. No differentiators between it and Wayland (Heck, you have to read the comments here even for a ment
Re: forwarding, vnc, xdmcp etc (Score:2)