Does anyone have insights into the possible motivation behind this? I'm an industry outsider, but I've been a bit baffled by the way that Canonical, Purism, and now Red Hat are seemingly doubling-down on Gnome and shifting away from KDE Plasma at precisely the moment that (going by the discourse within my Linux infosphere, at least) Gnome is approaching a crisis point, stripping away features as the debt comes due for bad design decisions made (and good, if painful, design fixes not implemented) years ago, whereas KDE Plasma is cruising, constantly adding refinements onto an already good foundation.
Or to put it differently: all the Linux experts I read and listen to (many of whom actually work at Canonical or Red Hat) are talking about how great KDE Plasma is, and how troubled Gnome Shell is, yet all the companies are rejecting KDE for Gnome.
What gives? Is this just part of the growing "corporatization" of Linux -- i.e. an investment of resources into a more corporate-controlled project, with an eye on the bottom-line and optimizing business-consumer support, rather than into one that's more decentralized in its development and individual-user targeted (and that would just draw resources from the former)?
I don't know, it's always sort of been like this for KDE. Even going back to some of the earlier releases, it was always "That one DE with the funny library license". Qt was eventually moved to LGPL, but when stuff (desktop Linux) was starting to take off, KDE 4 came out and it was awful at first release. I vaguely remember Linus dumping on how bad it was which sure didn't help things.
Ubuntu got popular super quick, which got the DE popular, and the entire community just sort of got behind it.
If you're creating productivity tools for your customers, you probably don't want to channel money into multiple projects that compete with each other. So, it's not too surprising that RH wants to focus on just Gnome.
Ubuntu came long after the "funny licensing thing" - Qt was licensed under GPL for the first time in v2.2, which was in 2000. So the switch happened in KDE 2.x timeframe. And then KDE 3.x (which was arguably the pinnacle of the "classic" KDE series - the one that was competing against Gnome 2.x) was GPL from the get go, and that was 2002. The initial release of Ubuntu was in 2004.
Ubuntu was intentionally based on GNOME – before Ubuntu and Canonical even had names – because of the GNOME community's attitude towards usability and integration. (I was there.)
KDE is the 2000s model of window management, reminiscent of Vista with its gaudy transparency and compositing. The modern way is tablet-oriented UIs like Windows 10 Metro UI, which Gnome copied pretty well. Also Gnome is now scriptable with JavaScript. JavaScript for UIs is where the future has been headed for a while.
> all the Linux experts I read and listen to (many of whom actually work at Canonical or Red Hat) are talking about how great KDE Plasma is, and how troubled Gnome Shell is
People who are happy about the current primary/default environment don't often go out of their way to talk about it. People who are happy about a non-default environment, or who are not happy with the default environment, talk about it more. It's a form of selection bias.
I have been a interested onlooker for > 10 years. As a programmer, KDE always seemed and still seems awesome and ahead of GNOME, API wise.
As a user, KDE always felt like an annoying blinkenlight kitchensink where everything could be modified. Gnome started out a little bit like that (with config files) but settled into a mediocre but unsurprising and thus stable and useful environment.
Before switching to MacOS I used Linux on my desktop for a decade.
KDE has always been the one with the better foundation, except that ... it never worked ok.
Every time I tried it, it had a ton of issues that would make me go into this rabbit hole, trying to fix it via endless configuration options and it never worked. And the migration from KDE 3 to KDE 4 was a complete clusterfuck.
Whereas Gnome always lacked configuration options, people always complained about its state, but it kept compatibility and it worked well out of the box.
The Linux desktop never took off due to immaturity. Microsoft did not abandon Windows, they didn’t rewrite from scratch, every release is basically an iteration of the previous one and Windows 10 for example is simply an iteration of Windows 7, which was XP with some lipstick on it. Same can be said of MacOS.
In open source however people just want to work on the new shiny all the time. As soon as something with more sex appeal comes along, people move to it.
The desktop Linux is freaking horrible and obviously what people have been doing in the last two decades doesn’t work. So yes, let’s corporatize it, let’s make it boring and reliable. Because people want a working desktop, not freaking widgets.
Speaking of desktop widgets, this is yet another example of immaturity. Do you know how many times per week I see my MacOS desktop? Never. My apps are always on and I restart my laptop like once per month, otherwise it’s in standby all the time. Such reliably is unheard of with the Linux desktop.
> My apps are always on and I restart my laptop like once per month, otherwise it’s in standby all the time. Such reliably is unheard of with the Linux desktop.
Bullshit. I'm at 42 days on a ThinkPad running Void Linux.
KDE has never worked fantastically on Ubuntu especially its long term support release. Ubuntu persistently ships out of date libraries with bugs that have in fact already been fixed in newer versions that wont be updated for months/years until the next long term stable release. KDE's releass aren't delivered in sync with ubuntu's calendars so it has always worked better with a more up to date distro.
Widgets on the desktop seem to be a pointless dead end that we keep revisiting in lots of different environments. Maybe faddish stuff like that is more interesting to work on than spit and polish. That said
"Microsoft did not abandon Windows, they didn’t rewrite from scratch,"
They actually did this more than once in more ways than one that you believe this is so is more of an indication of their obsessive focus on backwards compatiability. Fro example Windows started out as an app running on dos, windows NT/XP was a completely different OS from 95 98. The underlying layers used to build the windows experience have changed multiple times. The ui has been given a substantial overhaul with vista/8. The way drivers worked changed a lot with vista.
"The desktop Linux is freaking horrible and obviously what people have been doing in the last two decades doesn’t work. So yes, let’s corporatize it, let’s make it boring and reliable. Because people want a working desktop, not freaking widgets."
As someone who has been using desktop linux for 15 years now I'm a little confused. KDE 3 was quite usable. KDE 4 was kind of annoying to use 4.0 - 4.3 which was about the begining of 08 through mid 09 but at the time kde + compiz was a thing and was honestly much better.
There are now actually a variety of simple and complete environments all of which work well. Of the ones I have currently sampled. Cinnamon, Plasma, i3wm, bspwm, mate all work well. I have 3 different linux distributions running in my house right now. All work well. The only one that was work to install or maintain is a derivative of gentoo which honestly has been less effort than running windows. What's horrible? We have an embarassment of riches. We have too many aweesome things to choose from.
The Linux desktop never took off because it lacked preinstalls. This could be due to immaturity, but at the time Microsoft was strong-arming OEMs, so it was unlikely to happen either way. Now it's too late.
I'd always had the same kind of buggy KDE experience whenever I experimented with it, so I used Gnome for years. And then I tried KDE Plasma... it works fantastically, at least for me, on the latest stable Ubuntu. So it's sad to see RH moving away. Gnome is such a dumpster fire right now (and has been since the inception of Gnome 3)
My Ubuntu Mate 16.04 often gets up into the several months of uptime before I find some reason to reboot, and none of that has something to do with Gnome/Mate
(interesting to compare with their KDE developers just below).
The whole reason everybody got stuck with systemd is because Gnome wouldn't run without it. Every distribution had a choice of getting rid of Gnome or moving to systemd.
I actually gave up Gnome several years ago when they decided to make it so that every window shared an IME. That would mean that I couldn't have a window that had Japanese input and a window that had Roman input at the same time. Red Hat developers were busy making patches to every IME that existed so that it would follow the grand vision of what they had decided for Gnome. The only hold out was FCITX. I think they eventually saw reason but there was a good year or so where Red Hat was bent on breaking my computer just because that's what the Gnome developers had decided was best for everybody.
Red Hat doubled down on Gnome a long time ago. They aren't interested in anything else.
Go down the list a bit and notice: "GNOME Shell - Primary development by Owen Taylor and many Red Hat developers and interaction designers" among many other core Gnome infrastructure.
> The whole reason everybody got stuck with systemd is because Gnome wouldn't run without it. Every distribution had a choice of getting rid of Gnome or moving to systemd.
I've always found it interesting that, in my opinion, Gnome has always felt better to use as a user, but KDE had consistently and vastly better user applications than the Gnome counterparts.
That said, I've come to love the layout of Gnome 3 (with many extensions, by default it's nearly unusable). However, its performance is awful. While it's better than it used to be, it's still a leaky mess. I for one hope KDE continues to exist and be supported for no reason other than having a viable alternative when I finally become tired of performance.
As a developer who looked into the two a few years back, I chose QT as it was much easier to work with than GTK. This was before QML, however, so I don't know if it still holds true.
In my experience using both for years, Gnome has always been more stable.
Multi-monitor, Hi-Dpi support, in particular, is properly broken in KDE (while it's only half broken on Gnome).
As for features, the titlebar integration used by a few applications (ala MacOS) is a nice touch.
Otherwise it's true that the KDE experience, especially around customization, seems more thought out. Gnome's UX is nonsensical in more than a few places. The problem with KDE has always been stability and reliability. Having to reboot your machine due to crashes even once a week is enough to justify putting up with Gnome's overall worst UX.
> Hi-Dpi support, in particular, is properly broken in KDE
Not true. I use KDE everyday on my primary computer (an HP Envy 17t laptop), and have so since June this year. It works excellently on my 4K display. (You even get to pick a non-integer scaling factor.)
Just FYI, I run Arch Linux, and am (consequently) always on the latest version of KDE.
Everything worked out of the box with KDE. Including my laptop’s various hot keys, sleep/wake, networking, etc.
KDE was quite incredible compared to the 1 kLOC xmonad that I’d been using in my previously laptop. Zero configuration. Everything working perfectly with no headache.
Does multi-monitor mixed-DPI work in Gnome? Because that's the only thing that I couldn't get working in KDE about 2 years ago. (The problem was most easily solved with a new monitor.)
I restart perhaps every six months due to some problem that could be KDE; I find it very reliable both at home and at work. Granted, both machines are desktops, I use a laptop so rarely that I usually shut it down after use.
> titlebar integration
This is personal preference, but that's a showstopper for me ever using Gnome (or Mac OS) for anything beyond web browsing. I use focus-follows-mouse, which is incompatible with a titlebar outside the relevant window.
This is RHEL. People buy support contracts for it because they want to run Oracle's database and other enterprise-y stuff on headless computers sitting in a rack somewhere.
As someone who uses RHEL (and also CentOS because we do our own support) at work all day long I'm surprised it had a supported desktop environment at all. I've spent tens of thousands of hours on logged into shells on RHEL systems and not once have I seen what the GUI looks like.
I'm sure RedHat to their delight has managed to find some clueless customer somewhere who's overpaying for RHEL as some GUI workstation environment, but that customer is going to be happy with GNOME.
I'd flip it around and ask how Gnome and KDE have existed as full-blown DE's for as long as they have considering the amount of work required.
I'd expect to see one DE for mainstream users, one DE for business, and a plethora of less well maintained options for power users (I don't use a DE at all).
The 'user' and 'business' DE I'd expect to coalesce into the same thing because they have fairly similar use cases.
The existence of multiple DE's strikes me as an anomaly, if anything.
KDE initially used QT which didn't have a satisfying license. Possibly incompatible, I'm not sure it was ever resolved entirely as to what the condition of things was. What was clear is that a corporate entity could close source QT and effectively end KDE. After years, multiple business transactions and such, QT was really open sourced all the way. By then RedHat had gone "all in" with BlueCurve and GNOME. I think as a result of that, and in spite of various standards about open desktop support, the commercial apps tend to be on GTK+; maybe not full GNOME apps, but Firefox and VMware and others have GTK+ in them and they will seamlessly use your distribution's custom skin... Ending support for KDE is just momentum. There could be a bit of a European vs North America thing too; I'm just speculating but take Suse vs RHEL, RHEL is tons more common in the US despite Suse's remarkably quality, there has been some questions about its future at times but it is a remarkably high quality platform with a really nice aesthetic about it but RHEL and Centos tend to dominate.
To be honest, I didn't know Redhat still supported KDE.
KDE is great, I think it's always been a bit more consistent, a bit more cohesive and it has a great set of well crafted applications.
With Redhat being bought by IBM and everything, the possible rumors about Canonical being in play, it sort of seems like it might be time for a new distribution to take hold. KDE is a great place to start, there is an awful lot of good stuff in there.
Meanwhile over on Ubuntu MATE, I'm enjoying gnome 2 and sticking with it. 100% disinterest in the cognitive load of learning any sort of new UI for the rest of my life.
> Does anyone have insights into the possible motivation behind this?
I consider the following as possible reasons:
1. GTK+ is completely (GNOME) community owned (QT is a company owned, which companies like to avoid)
2. QT is LGPLv3+/GPLv3+. This, I assume, is to encourage buying commercial license than promoting free software. GTK+ is LGPLv2.1+, a lot of individual/small team developers prefer [L]GPLv2 over [LA]GPLv3.
3. Accessibility is pretty bad in GNU/Linux in general. But if there is one, the only choice is AT-SPI[0]. It's a GNOME project. I think KDE is a bit behind (I'm not sure though).
4. GNOME, GTK+ are GNU projects (Or many people prefer GNOME because it's a GNU project). I have seen many people in our Free Software User Group using GNOME just because of this.
> Gnome is approaching a crisis point, stripping away features as the debt comes due for bad design decisions made
We have seen a lot of software pieces whose development stalls, just because the code complexity increases way beyond some limit, or the main developer quits. Many (vocal) users prefer well configurable software pieces - but it comes with a high price. Many implement every possible feature to look the application c00l. But eventually as we gain experience, we may eventually abandon the project just because it's bloated (code wise), or even re-write it with less features, so that the code is more maintainable. If we look the code of many abandoned projects, we can see how bloated the code is, with many hacks, and bad designs.
I think GNOME is in the path of porting their applications to maintainable pieces, and eventually adding features as the code gets better.
> 1. GTK+ is completely (GNOME) community owned (QT is a company owned, which companies like to avoid)
If avoiding commercial vendors is what you want, do not look at the emails in the GTK commit messages :-).
Qt is covered by an agreement with the KDE Free Qt Foundation which ensures that even if the company decides to stop developing it (even in case of bankruptcy or acquisition) or to stop releasing it as open source, the Foundation can release it under an open source license.
It's in no worse shape than GTK+ in this regard, except it has more paid developers.
> 2. QT is LGPLv3+/GPLv3+. This, I assume, is to encourage buying commercial license than promoting free software. GTK+ is LGPLv2.1+, a lot of individual/small team developers prefer [L]GPLv2 over [LA]GPLv3.
This is, legit, the other way 'round. Large companies are the ones preferring GPLv2 over GPLv3 because of the tivoization clause. In the services/consulting industry, ensuring that a BSP is free of (L)GPLv3 components is basically a standard statement in SoWs.
Small teams and independent developers sometimes stick with GPLv2 because it has fewer legal pitfalls and they don't mind commercial attention, but it's not in your favour :-).
The code complexity of Gnome has been steadily decreasing, not increasing in the last 7-8 years. The impending "crisis" has nothing to do with code complexity and everything to do with how design and implementation decisions were made (and, of course, with many of those decisions).
Who cares? Let's just pick one - any - and standardise on that one so we can maybe move towards a consistent look and feel. They're all sort of the same; just different enough to be annoying and frighten any occasional/non-technical is away. Gnome is good enough? That'll do, let's use that then.
I think the poster's point is that they're not. One has apparently been well designed to be able to scale to new features, and the other one appears not to have been.
People like options, especially the Linux/FOSS people. You could come to a consensus about something but it'll only go so far until some developer decides to fork it and make their own project :)
If you go one step further, who cares about kernels and shells? Let's just standardise on one (NT) so we can maybe move towards actually getting work done. Windows is good enough? That'll do, let's use that then. /s
> Or to put it differently: all the Linux experts I read and listen to (many of whom actually work at Canonical or Red Hat) are talking about how great KDE Plasma is, and how troubled Gnome Shell is, yet all the companies are rejecting KDE for Gnome.
> What gives?
Simple question to answer: why do so many devs work on OS X machines?
People value "it works without me thinking about it" over "it has features".
However, the people who value "features" are really loud.
The only thing most people use computers for is sharing pictures and videos. And what are the things that Linux sucks at making work out of the box: wireless, audio and video.
KDE could win the battle simply by making those work--everysingletime.
OS X does it. And look at how many developers use it.
It's a different playing field: OSX has to support a small handful of chipsets, and if it doesn't work then manufacturers will step in. Linux has to support a boatload of crap silicon, and if anything doesn't work, manufacturers will just shrug.
Add to it that every Linux distribution has its idiosincratic approach to low-level subsystems, and you have a recipe for problems if you work anywhere higher up the stack.
But don't worry, eventually the problem will be solved the other way around: there will be a systemd-DE, and everyone will use that. /s
I don't know any numbers, but my guess is that the majority of Red Hat's GNU/Linux subscriptions are for the headless server version.
I'm sure they have some customers for the Workstation edition, but I've always felt that the Desktop features were done more about of employee desire than top down mandates.
>> What gives? Is this just part of the growing "corporatization" of Linux -- i.e. an investment of resources into a more corporate-controlled project
I always felt KDE was more of a corporate project. It's also been my impression that KDE comes with a lot more dependencies. IMHO GTK is a decent toolkit with lower complexity - it's GNOME shell that I really don't like but I live with it.
KDE was hardly ever "corporate", but it was always somewhat controlled / guided by the company behind Qt (Trolltech, originally); for the simple reason that most of the top KDE hackers were top Qt hackers and hence, in most cases, employees of Qt's owners. In many ways, getting involved with KDE was a good way to audition for a job at Trolltech/Nokia/Digia/whatever-it-is-now.
Add to it that KDE saw some significant early deployments in the European public sector, which resulted in some particular attention given to collaboration features (imap, calendaring etc). That might have looked corporate to some, but it was just a necessity for bootstrapped QT-centric businesses trying to take on the office giants like Microsoft and IBM.
Gnome was built on GTK and GObject, libraries that have always had well defined licensing.
KDE was built on Qt, a framework that did not always have the licensing it has today.
GTK and GObject are C libraries, and many constructs are easy to reason about. Qt is C++ based and there is a LOT going on and the meta compiler is insane.
This an important and pertinent question. I’ve often wondered the same so I thank you for asking it. I find QT UI applications just a bit more consistent than GNOME ones. It’s a subjective impression...
Well, it is very simple: Linux expert users do not represent the user base that RHEL is interested in.
RHEL is interested about that one guy or woman that actually decides(to pay them) and has control of the company. This person wants control overall of what people do with (her company) computers.
Desktops like Gnome have a very limited set of options, great for companies. More stable, less problems, people focusing more on work to do at the company.
I used to program KDE widgets for things like displaying the tides, the lunar phases, sunset times and so on. It was a cool thing to do and useful if I wanted to go fishing or to the beach. Marginally useful for work.
Working you need something that works, with the least amount of work and money possible. KDE is not designed for that.
> What gives? Is this just part of the growing "corporatization" of Linux
I would argue that this is the case, and if it's not the case then mostly because it's not becoming more corporized but because it always was.
I would argue that there are three popular reasons for open source development, that there is a huge difference in user impact between these, that corporate interests have been the main driver behind open source, that having ones own preferred solution allowed to take profit from the competition, and now that the profits are dwindling distinction between competitors becomes less important than saving costs, and therefore everybody focusses on whatever appears to be as the winner in the current topic area.
Reason 1 for open source development, and probably the original reason, is art. People wanted to scratch an itch, they themself felt. They don't care much about gaining any additional advantage so they share the whole thing and not just the binaries, maybe by doing so gaining some recognition by other smart people. These artist developers usually hate the normal user and despise the disrespect these bring towards their piece of art. They often don't care about the trivial problems the user has, like the need for a complex yet intuitive UI. So users will often refrain from using these tools. Making them mostly unimportant at large, if not someone else comes along and uses the knowledge gained from it for another category of open source.
Reason 2 for open source development is the goal of social and political influence, which drives a person or group to develop tools to fascilitate communication between supporters and fascilitate spread of the culture. One example I see here is the scuttlebut network, which if you analyse closely, is not really good for everybody but mostly used by a group of New Zealend developers to spread their ideas of off-grid lifesstyle and something they call "solar punk". This is by no means criticism. It's a great tool/protocol and the ideas they spread with it also don't seem harmful to me. But without the desire to influence the political landscape they probably wouldn't invest that amount of time and energy to build their social network. This kind of project can gain quite a lot of influence on users at large. For instance all these 2chan,4chan,8chan imageboards might rival professional social networks in size and usability. But usually it's limited to people who support the political goals or at least can accept them. For instance it's unlikely that a group of feminist game developers would choose 8chan as their platform after it became a platform for the anti-feminist gamergaters.
Last but not least stand the group of open source project which directly or indirectly are for profit. For instance Linus Torvalds is one of the best earning people on this planet. Multiple companies have formed around linux distributions. And most of the commits nowadays come from people who are paid to do them. As said as it may sound, this is the most influential group. It's simply a fact that user-focussed development can be a daunting task and most people wouldn't do it if nobody paid them for it. So the development is also built around market dynamics. When cool UIs gain user attention most distros will focus their resources on one desktop manager that separates them from the rest, so that users who like these distinctive features will come to them. And when profits shrink they will look for whatever desktop manager is the current winner and jump on that band wagon, so they can reduce their development investments and focus these on other topics.
If we can agree on these three categories I think it's clear that the last one is the one that gains the most users, and therefore will also influence open source in general the most. Also it might have never been different from that, and as long as the getting is good it's also not to the users disadvantage.
To clarify, this is related to Red Hat Enterprise Linux the distro, not Red Hat the company. Fedora 29 (released a few days ago) has a KDE spin, and the project has made no mention of discontinuing that. https://spins.fedoraproject.org/kde/download/index.html
Full disclosure: I work for Red Hat (but not on RHEL or Fedora).
Somewhat off topic, but what's the reaction at Red Hat been about the IBM deal? I live in North Carolina and have been looking for jobs lately; I had been seriously considering Red Hat before the deal was announced, but I'm not sure what to think. I was hoping to go somewhere that offered more stability than a startup, but with the uncertainty around what IBM's plans are, I'm not sure that's the case for Red Hat at the moment.
Meanwhile KDE5 is, IMHO, the only reasonable modern classic desktop for Linux (XFCE, LXDE, LXQt and Mate are also classic desktops but these are not really modern i.e. they are more Win95-like while KDE5 is more Win7/Unity-like and more), every else is a way too "special" and GNOME3 is the most weird one. From what I know about GNOME3 (from experience of attempting to use it and some reading/watching trying to "get it") it would be great to use on a tablet but looks and feels rather clumsy on a laptop.
As someone who landed on Gnome 3 quite a few years ago and hasn't looked back, I'm really curious: why do you feel it would be great to use on a tablet?
To me, it doesn't seem that great on a tablet. A major part of what makes Gnome 3 amazing for me is how easily I can navigate between workspaces or move windows between my screens and workspaces using keyboard shortcuts, but tablets don't even have physical keyboards. The whole "switch workspaces instead of windows" workflow of Gnome 3 is great on desktops where people multi-task and have complex contexts on each workspace, but to me it seems that this workflow would be a lot less useful with a single screen, and even more useless with a tiny tablet screen.
Aren't workspaces really useful when your physical screen space is limited (when it's not limited much I can hardly get it why won't you just arrange everything you need on your physical screens and just minimize/unminimize apps that you don't need all the time, KDE does this with just WindowsButton+number and it still has activities too although I never found this feature useful on a modern PC with 2 reasonably big screens)? Aren't huge window decorations great for touch screens? As for lack of keyboard - introducing a panel of touchscreen buttons for all the necessary shortcuts or using gestures may probably do the job.
BTW, for those who had only tried KDE years ago (i.e. KDE4 and/or older and/or earliest KDE5 releases) and didn't really like it (e.g. default KDE look&feel had always felt mediocre to me and reliability of KDE4 and early KDE5 was bad), today KDE5 looks and feels much much better (to me at least, I dunno about default themes/settings in other distros but in Manjaro it's delicious and makes an awesome replacement for Unity (R.I.P.) after some tweaking).
I find it odd there's so much emotion about KDE vs Gnome in RHEl/CentOS. RHEL is an 'enterprise' distro, which means in theory professionals are using it inside of their application of choice, and almost never interacting in a meaningful way with the DE. They need almost no customization and only the simplest file manipulation tools, and the ability to open the programs they do their work in. If you want to use it at home, you can just install KDE yourself, why expect rhel to support it?
Source: Former sysadmin for ~5k CentOS desktop users (using gnome)
I really like Gnome and think KDE is way too bloated visually. On the other hand, Gnome is literally a dumpster fire of resource use. On some of my machines it would instantly shoot to 30-60% cpu. It uses a metric ton of RAM.
That's why I recommend sway! the amazing wayland tiling windows manager. It's nice to look at, easy to use, and it uses like less hz than my wrist watch.
Anything other than Gnome or KDE will use less than half the resources. XFCE, Mate, LXDE/LXQT, etc. while maintaining the whole "environment" part of "desktop environment".
There's a huge difference between a tiling WM and a full-blown DE, and people who want the Gnome or KDE experience should absolutely not install sway or any other tiling WM.
DEs should make it easier to use tiling as a feature rather than having to go all-in with something like i3 or awesome.
I use xfce, and it's pretty easy to add sensible keybindings for snap-window-{up, down, left, right, top-left, top-right, bottom-left, bottom-right, fullscreen}. That's usually all I want, although occasionally it would be nice to be able to quickly split by thirds.
This still feels super snappy and productive, and you still get people asking you "whoa, how did you do that!?" But you can also overlap windows without changing modes, and the usual window manipulators are still there when they make sense, or when someone else wants to use your computer.
You shouldn't have to throw away the rest of the "normal" DE and relearn how to interact with your computer to get tiling. It should just be a secret superpower for people who know that it's there. I've never understood why it's always one or the other.
Most likely the fact that they’re running ancient kernel versions (RHEl 7.6, released Oct 30 2018 uses Linux 3.10, which was released in June 2013) , and have to support them for 10-20 years.
RHEL 6 uses kernel version 2.6.32, released in 2011, and will be supported until 2020.
With a moving target like Btrfs, which may have plenty of showstopper bugs left in it yet, it’s just not feasible to keep backporting these new features to a an old kernel version.
Once Btrfs stops getting major patches, and is considered stable, I guess it could eventually make its way back.
While true, there's also the fact that they don't have any btrfs engineers anymore (from what I've heard). The ones they had have all left for Facebook and other companies that use btrfs in production.
Redhat have added new filesystem support after a .0 release. For example, they added XFS to the kernel (without xfsprogs) in RHEL 5.2 and added the tools to the base repo in 5.4, then officially supported it in 5.6.
From reports I've read by insiders, it's mainly due to the fact that all the filesystem developers in RH are XFS-focused, and everybody who's XFS-focused isn't interested in learning enough btrfs to keep their end up-to-date. So when the last guy who did btrfs left, they were effectively forced to end their support for it.
Sendmail is deprecated. That was the first Unix service I ever learned to configure. This is kind of the same feeling I got when I heard music I grew up listening to on the classic rock station.
Readit News
Or to put it differently: all the Linux experts I read and listen to (many of whom actually work at Canonical or Red Hat) are talking about how great KDE Plasma is, and how troubled Gnome Shell is, yet all the companies are rejecting KDE for Gnome.
What gives? Is this just part of the growing "corporatization" of Linux -- i.e. an investment of resources into a more corporate-controlled project, with an eye on the bottom-line and optimizing business-consumer support, rather than into one that's more decentralized in its development and individual-user targeted (and that would just draw resources from the former)?
Then Ubuntu came out (well, before KDE 4, but during the funny licensing thing), which used Gnome because Debian used Gnome. Debian used Gnome because Qt had the QPL licensing: https://wiki.debian.org/DFSGLicenses#Q_Public_License_.28QPL...
Ubuntu got popular super quick, which got the DE popular, and the entire community just sort of got behind it.
If you're creating productivity tools for your customers, you probably don't want to channel money into multiple projects that compete with each other. So, it's not too surprising that RH wants to focus on just Gnome.
People who are happy about the current primary/default environment don't often go out of their way to talk about it. People who are happy about a non-default environment, or who are not happy with the default environment, talk about it more. It's a form of selection bias.
- how I get tray icons working (tray icons were removed)
- GNOME won't start (an extension incompatibility can make the DE crash)
- lag when some animation happens
- how can I make the theme to use less padding (some devs are considering removing theming support completely )
This are complaints for actual GNOME users that update to latest version and have unpleasantness surprises.
As a user, KDE always felt like an annoying blinkenlight kitchensink where everything could be modified. Gnome started out a little bit like that (with config files) but settled into a mediocre but unsurprising and thus stable and useful environment.
Deleted Comment
KDE has always been the one with the better foundation, except that ... it never worked ok.
Every time I tried it, it had a ton of issues that would make me go into this rabbit hole, trying to fix it via endless configuration options and it never worked. And the migration from KDE 3 to KDE 4 was a complete clusterfuck.
Whereas Gnome always lacked configuration options, people always complained about its state, but it kept compatibility and it worked well out of the box.
The Linux desktop never took off due to immaturity. Microsoft did not abandon Windows, they didn’t rewrite from scratch, every release is basically an iteration of the previous one and Windows 10 for example is simply an iteration of Windows 7, which was XP with some lipstick on it. Same can be said of MacOS.
In open source however people just want to work on the new shiny all the time. As soon as something with more sex appeal comes along, people move to it.
The desktop Linux is freaking horrible and obviously what people have been doing in the last two decades doesn’t work. So yes, let’s corporatize it, let’s make it boring and reliable. Because people want a working desktop, not freaking widgets.
Speaking of desktop widgets, this is yet another example of immaturity. Do you know how many times per week I see my MacOS desktop? Never. My apps are always on and I restart my laptop like once per month, otherwise it’s in standby all the time. Such reliably is unheard of with the Linux desktop.
Bullshit. I'm at 42 days on a ThinkPad running Void Linux.
> 19:19:22 up 42 days, 19:53, 1 user, load average: 2.37, 0.98, 0.56
I run my KDE desktop 24/7 in very much the way you are describing and it works very reliably.
Widgets on the desktop seem to be a pointless dead end that we keep revisiting in lots of different environments. Maybe faddish stuff like that is more interesting to work on than spit and polish. That said
"Microsoft did not abandon Windows, they didn’t rewrite from scratch,"
They actually did this more than once in more ways than one that you believe this is so is more of an indication of their obsessive focus on backwards compatiability. Fro example Windows started out as an app running on dos, windows NT/XP was a completely different OS from 95 98. The underlying layers used to build the windows experience have changed multiple times. The ui has been given a substantial overhaul with vista/8. The way drivers worked changed a lot with vista.
"The desktop Linux is freaking horrible and obviously what people have been doing in the last two decades doesn’t work. So yes, let’s corporatize it, let’s make it boring and reliable. Because people want a working desktop, not freaking widgets."
As someone who has been using desktop linux for 15 years now I'm a little confused. KDE 3 was quite usable. KDE 4 was kind of annoying to use 4.0 - 4.3 which was about the begining of 08 through mid 09 but at the time kde + compiz was a thing and was honestly much better.
There are now actually a variety of simple and complete environments all of which work well. Of the ones I have currently sampled. Cinnamon, Plasma, i3wm, bspwm, mate all work well. I have 3 different linux distributions running in my house right now. All work well. The only one that was work to install or maintain is a derivative of gentoo which honestly has been less effort than running windows. What's horrible? We have an embarassment of riches. We have too many aweesome things to choose from.
https://fedoraproject.org/wiki/Red_Hat_contributions#GNOME_d...
(interesting to compare with their KDE developers just below).
The whole reason everybody got stuck with systemd is because Gnome wouldn't run without it. Every distribution had a choice of getting rid of Gnome or moving to systemd.
I actually gave up Gnome several years ago when they decided to make it so that every window shared an IME. That would mean that I couldn't have a window that had Japanese input and a window that had Roman input at the same time. Red Hat developers were busy making patches to every IME that existed so that it would follow the grand vision of what they had decided for Gnome. The only hold out was FCITX. I think they eventually saw reason but there was a good year or so where Red Hat was bent on breaking my computer just because that's what the Gnome developers had decided was best for everybody.
Red Hat doubled down on Gnome a long time ago. They aren't interested in anything else.
Edit: Also of note - https://fedoraproject.org/wiki/Red_Hat_contributions#Desktop...
Go down the list a bit and notice: "GNOME Shell - Primary development by Owen Taylor and many Red Hat developers and interaction designers" among many other core Gnome infrastructure.
They haven't made it easy to run GNOME Shell without systemd, but people have got it working in Gentoo, Void etc., see, for instance: https://dm.reddit.com/r/linux/comments/66ass2/update_void_li...
(That said, I don't run GNOME Shell on any of my machines.)
That said, I've come to love the layout of Gnome 3 (with many extensions, by default it's nearly unusable). However, its performance is awful. While it's better than it used to be, it's still a leaky mess. I for one hope KDE continues to exist and be supported for no reason other than having a viable alternative when I finally become tired of performance.
As a developer who looked into the two a few years back, I chose QT as it was much easier to work with than GTK. This was before QML, however, so I don't know if it still holds true.
I do not understand how GNOME feels different since you can setup KDE to behave like you want,
As for features, the titlebar integration used by a few applications (ala MacOS) is a nice touch.
Otherwise it's true that the KDE experience, especially around customization, seems more thought out. Gnome's UX is nonsensical in more than a few places. The problem with KDE has always been stability and reliability. Having to reboot your machine due to crashes even once a week is enough to justify putting up with Gnome's overall worst UX.
Not true. I use KDE everyday on my primary computer (an HP Envy 17t laptop), and have so since June this year. It works excellently on my 4K display. (You even get to pick a non-integer scaling factor.)
Just FYI, I run Arch Linux, and am (consequently) always on the latest version of KDE.
Everything worked out of the box with KDE. Including my laptop’s various hot keys, sleep/wake, networking, etc.
KDE was quite incredible compared to the 1 kLOC xmonad that I’d been using in my previously laptop. Zero configuration. Everything working perfectly with no headache.
I restart perhaps every six months due to some problem that could be KDE; I find it very reliable both at home and at work. Granted, both machines are desktops, I use a laptop so rarely that I usually shut it down after use.
> titlebar integration
This is personal preference, but that's a showstopper for me ever using Gnome (or Mac OS) for anything beyond web browsing. I use focus-follows-mouse, which is incompatible with a titlebar outside the relevant window.
As someone who uses RHEL (and also CentOS because we do our own support) at work all day long I'm surprised it had a supported desktop environment at all. I've spent tens of thousands of hours on logged into shells on RHEL systems and not once have I seen what the GUI looks like.
I'm sure RedHat to their delight has managed to find some clueless customer somewhere who's overpaying for RHEL as some GUI workstation environment, but that customer is going to be happy with GNOME.
I'd expect to see one DE for mainstream users, one DE for business, and a plethora of less well maintained options for power users (I don't use a DE at all).
The 'user' and 'business' DE I'd expect to coalesce into the same thing because they have fairly similar use cases.
The existence of multiple DE's strikes me as an anomaly, if anything.
Well, I’ve got good news for you...
Just curious, what do you use?
KDE initially used QT which didn't have a satisfying license. Possibly incompatible, I'm not sure it was ever resolved entirely as to what the condition of things was. What was clear is that a corporate entity could close source QT and effectively end KDE. After years, multiple business transactions and such, QT was really open sourced all the way. By then RedHat had gone "all in" with BlueCurve and GNOME. I think as a result of that, and in spite of various standards about open desktop support, the commercial apps tend to be on GTK+; maybe not full GNOME apps, but Firefox and VMware and others have GTK+ in them and they will seamlessly use your distribution's custom skin... Ending support for KDE is just momentum. There could be a bit of a European vs North America thing too; I'm just speculating but take Suse vs RHEL, RHEL is tons more common in the US despite Suse's remarkably quality, there has been some questions about its future at times but it is a remarkably high quality platform with a really nice aesthetic about it but RHEL and Centos tend to dominate.
To be honest, I didn't know Redhat still supported KDE.
KDE is great, I think it's always been a bit more consistent, a bit more cohesive and it has a great set of well crafted applications.
With Redhat being bought by IBM and everything, the possible rumors about Canonical being in play, it sort of seems like it might be time for a new distribution to take hold. KDE is a great place to start, there is an awful lot of good stuff in there.
I don't care about all the tweaks and config options as much as I used to 15 years ago when I started using linux.
To each his own.
Deleted Comment
I consider the following as possible reasons:
1. GTK+ is completely (GNOME) community owned (QT is a company owned, which companies like to avoid)
2. QT is LGPLv3+/GPLv3+. This, I assume, is to encourage buying commercial license than promoting free software. GTK+ is LGPLv2.1+, a lot of individual/small team developers prefer [L]GPLv2 over [LA]GPLv3.
3. Accessibility is pretty bad in GNU/Linux in general. But if there is one, the only choice is AT-SPI[0]. It's a GNOME project. I think KDE is a bit behind (I'm not sure though).
4. GNOME, GTK+ are GNU projects (Or many people prefer GNOME because it's a GNU project). I have seen many people in our Free Software User Group using GNOME just because of this.
> Gnome is approaching a crisis point, stripping away features as the debt comes due for bad design decisions made
We have seen a lot of software pieces whose development stalls, just because the code complexity increases way beyond some limit, or the main developer quits. Many (vocal) users prefer well configurable software pieces - but it comes with a high price. Many implement every possible feature to look the application c00l. But eventually as we gain experience, we may eventually abandon the project just because it's bloated (code wise), or even re-write it with less features, so that the code is more maintainable. If we look the code of many abandoned projects, we can see how bloated the code is, with many hacks, and bad designs.
I think GNOME is in the path of porting their applications to maintainable pieces, and eventually adding features as the code gets better.
[0] https://en.wikipedia.org/wiki/Assistive_Technology_Service_P...
If avoiding commercial vendors is what you want, do not look at the emails in the GTK commit messages :-).
Qt is covered by an agreement with the KDE Free Qt Foundation which ensures that even if the company decides to stop developing it (even in case of bankruptcy or acquisition) or to stop releasing it as open source, the Foundation can release it under an open source license.
It's in no worse shape than GTK+ in this regard, except it has more paid developers.
> 2. QT is LGPLv3+/GPLv3+. This, I assume, is to encourage buying commercial license than promoting free software. GTK+ is LGPLv2.1+, a lot of individual/small team developers prefer [L]GPLv2 over [LA]GPLv3.
This is, legit, the other way 'round. Large companies are the ones preferring GPLv2 over GPLv3 because of the tivoization clause. In the services/consulting industry, ensuring that a BSP is free of (L)GPLv3 components is basically a standard statement in SoWs.
Small teams and independent developers sometimes stick with GPLv2 because it has fewer legal pitfalls and they don't mind commercial attention, but it's not in your favour :-).
The code complexity of Gnome has been steadily decreasing, not increasing in the last 7-8 years. The impending "crisis" has nothing to do with code complexity and everything to do with how design and implementation decisions were made (and, of course, with many of those decisions).
I think the poster's point is that they're not. One has apparently been well designed to be able to scale to new features, and the other one appears not to have been.
Deleted Comment
> What gives?
Simple question to answer: why do so many devs work on OS X machines?
People value "it works without me thinking about it" over "it has features".
However, the people who value "features" are really loud.
The only thing most people use computers for is sharing pictures and videos. And what are the things that Linux sucks at making work out of the box: wireless, audio and video.
KDE could win the battle simply by making those work--every single time.
OS X does it. And look at how many developers use it.
Add to it that every Linux distribution has its idiosincratic approach to low-level subsystems, and you have a recipe for problems if you work anywhere higher up the stack.
But don't worry, eventually the problem will be solved the other way around: there will be a systemd-DE, and everyone will use that. /s
Too bad the 3 are OS problems, and the DE has no way to fix them.
I always felt KDE was more of a corporate project. It's also been my impression that KDE comes with a lot more dependencies. IMHO GTK is a decent toolkit with lower complexity - it's GNOME shell that I really don't like but I live with it.
Add to it that KDE saw some significant early deployments in the European public sector, which resulted in some particular attention given to collaboration features (imap, calendaring etc). That might have looked corporate to some, but it was just a necessity for bootstrapped QT-centric businesses trying to take on the office giants like Microsoft and IBM.
KDE was built on Qt, a framework that did not always have the licensing it has today.
GTK and GObject are C libraries, and many constructs are easy to reason about. Qt is C++ based and there is a LOT going on and the meta compiler is insane.
RHEL is interested about that one guy or woman that actually decides(to pay them) and has control of the company. This person wants control overall of what people do with (her company) computers.
Desktops like Gnome have a very limited set of options, great for companies. More stable, less problems, people focusing more on work to do at the company.
I used to program KDE widgets for things like displaying the tides, the lunar phases, sunset times and so on. It was a cool thing to do and useful if I wanted to go fishing or to the beach. Marginally useful for work.
Working you need something that works, with the least amount of work and money possible. KDE is not designed for that.
Instead of configuring things, good point.
Deleted Comment
I would argue that this is the case, and if it's not the case then mostly because it's not becoming more corporized but because it always was.
I would argue that there are three popular reasons for open source development, that there is a huge difference in user impact between these, that corporate interests have been the main driver behind open source, that having ones own preferred solution allowed to take profit from the competition, and now that the profits are dwindling distinction between competitors becomes less important than saving costs, and therefore everybody focusses on whatever appears to be as the winner in the current topic area.
Reason 1 for open source development, and probably the original reason, is art. People wanted to scratch an itch, they themself felt. They don't care much about gaining any additional advantage so they share the whole thing and not just the binaries, maybe by doing so gaining some recognition by other smart people. These artist developers usually hate the normal user and despise the disrespect these bring towards their piece of art. They often don't care about the trivial problems the user has, like the need for a complex yet intuitive UI. So users will often refrain from using these tools. Making them mostly unimportant at large, if not someone else comes along and uses the knowledge gained from it for another category of open source.
Reason 2 for open source development is the goal of social and political influence, which drives a person or group to develop tools to fascilitate communication between supporters and fascilitate spread of the culture. One example I see here is the scuttlebut network, which if you analyse closely, is not really good for everybody but mostly used by a group of New Zealend developers to spread their ideas of off-grid lifesstyle and something they call "solar punk". This is by no means criticism. It's a great tool/protocol and the ideas they spread with it also don't seem harmful to me. But without the desire to influence the political landscape they probably wouldn't invest that amount of time and energy to build their social network. This kind of project can gain quite a lot of influence on users at large. For instance all these 2chan,4chan,8chan imageboards might rival professional social networks in size and usability. But usually it's limited to people who support the political goals or at least can accept them. For instance it's unlikely that a group of feminist game developers would choose 8chan as their platform after it became a platform for the anti-feminist gamergaters.
Last but not least stand the group of open source project which directly or indirectly are for profit. For instance Linus Torvalds is one of the best earning people on this planet. Multiple companies have formed around linux distributions. And most of the commits nowadays come from people who are paid to do them. As said as it may sound, this is the most influential group. It's simply a fact that user-focussed development can be a daunting task and most people wouldn't do it if nobody paid them for it. So the development is also built around market dynamics. When cool UIs gain user attention most distros will focus their resources on one desktop manager that separates them from the rest, so that users who like these distinctive features will come to them. And when profits shrink they will look for whatever desktop manager is the current winner and jump on that band wagon, so they can reduce their development investments and focus these on other topics.
If we can agree on these three categories I think it's clear that the last one is the one that gains the most users, and therefore will also influence open source in general the most. Also it might have never been different from that, and as long as the getting is good it's also not to the users disadvantage.
Full disclosure: I work for Red Hat (but not on RHEL or Fedora).
...yet.
To me, it doesn't seem that great on a tablet. A major part of what makes Gnome 3 amazing for me is how easily I can navigate between workspaces or move windows between my screens and workspaces using keyboard shortcuts, but tablets don't even have physical keyboards. The whole "switch workspaces instead of windows" workflow of Gnome 3 is great on desktops where people multi-task and have complex contexts on each workspace, but to me it seems that this workflow would be a lot less useful with a single screen, and even more useless with a tiny tablet screen.
Source: Former sysadmin for ~5k CentOS desktop users (using gnome)
That's why I recommend sway! the amazing wayland tiling windows manager. It's nice to look at, easy to use, and it uses like less hz than my wrist watch.
There's a huge difference between a tiling WM and a full-blown DE, and people who want the Gnome or KDE experience should absolutely not install sway or any other tiling WM.
Signed, a tiling WM user.
I use xfce, and it's pretty easy to add sensible keybindings for snap-window-{up, down, left, right, top-left, top-right, bottom-left, bottom-right, fullscreen}. That's usually all I want, although occasionally it would be nice to be able to quickly split by thirds.
This still feels super snappy and productive, and you still get people asking you "whoa, how did you do that!?" But you can also overlap windows without changing modes, and the usual window manipulators are still there when they make sense, or when someone else wants to use your computer.
You shouldn't have to throw away the rest of the "normal" DE and relearn how to interact with your computer to get tiling. It should just be a secret superpower for people who know that it's there. I've never understood why it's always one or the other.
Deleted Comment
RHEL 6 uses kernel version 2.6.32, released in 2011, and will be supported until 2020.
With a moving target like Btrfs, which may have plenty of showstopper bugs left in it yet, it’s just not feasible to keep backporting these new features to a an old kernel version.
Once Btrfs stops getting major patches, and is considered stable, I guess it could eventually make its way back.