With friends likr Gnome, you don’t need enemies anymore. If gnome were to collaborate with others, there would much less necessity for forks (such as in the GTK2 days). But instead, the close any feature request minimally differing their pristine idea of what a desktop should look like, shut down any other discussion about the topic, and god forbid if you mention you want customization like themes or, gasp, tray icons.
Sure, this may come off polemic, but I very much think that gnome is largely to blame for the fragmentation of the Linux desktop.
Redhat. They've made some power-plays (including with Gnome) and have been successful at it.
Ubuntu tried to do the same more than once, but fell so hard on their face every time that it can be hard to tell that they were trying to do the same thing.
The only "fragmentation" GNOME directly caused was MATE being forked out of GNOME 2. Fragmentation is just the natural evolution of FOSS because there isn't a monopoly shoving license terms down the throat of every customer. GNOME's philosophy is kind of restrictive but they are not the one to blame for the fragmentation of the Linux desktop especially considering DEs and distributions are SEPARATE things and many distributions provide multiple DE options bundled at install.
If GNOME did not exist, KDE would have been commercial ( Qt at least ). To prevent that, an alternative to GNOME would have arisen.
If KDE had not started with this flaw, I would agree with you.
For the most part, the Linux Desktop has been a majority using some variation of GNOME with a dedicated minority using KDE.
I have not used “GNOME” since the 2.x days but I have always been in the GNOME world. Even XFCE is really another base for GNOME.
I have not used KDE since the 2.x days either but current Plasma is pretty good and I do not mind it. I put Big Linux on a laptop and have not taken its version on KDE off yet.
> If GNOME did not exist, KDE would have been commercial ( Qt at least ). To prevent that, an alternative to GNOME would have arisen.
KDE wouldn't be commercial.
Qt didn't have the ideal license back than but you could use it free.
At the point KDE would be the de facto standard Unix desktop there would have been enough power behind that to arrive in the end at the exact same compromise regarding Qt that we have now, I'm quite sure.
There was no need to split the Unix desktop forever, and cause infinite pain as a result, by invoking the nuclear option, which the Gnome people did!
And given the further background story of one of the most prominent people behind that whole story you can even construct a reasonably believable "conspiracy theory" around the things that happened… ;-)
Also there are quite some other DEs. Only that there are just two big ones. If one of the big ones would not exist there would be a default one with the greatest market share.
Normal people don't reinvent the wheel when there is no need for that…
There was no need to split KDE. And Gnome exist solely for political reasons. Like I said, look at the history.
There wouldn't be a single standard desktop with >80% usage regardless of whether GNOME existed, and KDE will merely be among one of many desktop environment options. That's just the nature of how things work in this space.
You’re talking about Gnome as if it’s an end onto itself. But it did not appear out of thin air just to disrupt KDE’s dominance.
Look at the history again. Look at the names on the emails and posts.
It was done by people, because they wanted it done. For “solely political reasons”? But that’s exactly what politics means: making sure people get what they want. As opposed to everyone having what you want.
You are talking as if those two DEs are the only two options in existence. They are not. Xfce started development in 1996 ahead og GNOME. And if the "political reasons" you talked about were related to licensing issues with Qt, then GNOME is not the one to blame for. Asking people to look at the history is only going to make them discover the truth you don't want to admit.
If you have to use words like "most violent" and "nuclear" to describe a valid & normal act of launching free & open source software that everyone had the freedom of choice regarding whether to use it or not, you need to seriously re-evaluate your personal priorities and worldview.
I use both GNOME and KDE daily for different use cases, and the fact that people like you who hold such views exist is downright disturbing.
KDE was not "THE UNIX DESKTOP"...it never was, not for a single UNIX. If anything Gnome (aka Java Desktop) was a Unix Desktop (SUN Solaris) and is it still for Oracle Solaris.
Gnome 3 can't be responsible for Mate, Cinnamon and Budgie adding 3 fragments to the desktop environment space instead of those 3 adding 1 by merging each other.
Gnome 3 brushed people the wrong way but blaming it for fragmentation when those 3 went away to do their 3 different things doesn't make sense.
Come on, MATE and Cinnamon ? To be consistent Cinnamon shouldn't exist and only MATE would make the cut if Gnome 3 is the problem.
This is long overdue. The Linux desktop needs a lot of polishing and i feel like we've hit a wall with productivity considering the gnome file picker has had the same issues with thumbnails since i was in highschool with Ubuntu 6.10.
If Rust helps us actually improve and change things, then fine.
It needs more than just polishing. It would great to see the community come to agreement on a modern permissions model so that you can do things like grant access to your camera/microphone/calendar/location/photos/etc... for ever, just this time, or never. I know it's being worked on and there are some competing systems, but it would be great for there to be a single API.
I'm on KDE right now (KDE Neon). Basic stuff like, installing Chrome still uses the GTK file picker, is an issue. Loading screen goes to wrong monitor and oriented wrong. Plugging in monitor sometimes makes both go black requiring hard reboot (I think I fixed this with a kernel flag...). Just overall polish. Although I do like Neon as a distro so far.
Is there a right amount of fragmentation? Or is monolithic control like Windows and MacOS better? Part of the promise of open source software is that it can be adapted and extended. Part of the promise of Linux is that the kernel and the DE are not completely dependent on each other. So it can hardly be surprising when people take these promises at face value and make things with them.
Windows and Mac also have usability issues and they are driven by two companies that don't allow break ups of their DEs.
I don't think that a FOSS community will ever converge on an amazing DE without part of it breaking apart and starting a new one, for the sake of building something even more amazing or failing on the way.
I'm fine with trying new things. In particular, I'm curious as to whether people have tried to build a desktop environment around webkit or similar--basically using the web stack instead of traditional GUI toolkits? Is there some obvious reason this is a bad idea (yeah, I know working in JS is unpleasant, but we have alternatives these days and frankly all of the native GUI toolkits for Linux are pretty unpleasant as well).
EDIT: I know ChromeOS exists, but that's a lot more than a desktop environment (can't even run a compiler natively for all intents and purposes iirc).
Could be wrong, but I'm pretty sure Gnome just embeds a JavaScript runtime (and probably not a particularly good one) as an embedded scripting language. It's not using a web engine (which would handle styling, layout, accessibility, etc in addition to running JS).
I think they are referring to the human resources part of the fragmentation issue: capable people working on yet another desktop environment.
The, perhaps unrealistic and naive, thinking behind this is that if these capable people combined their efforts into one desktop environment, we'd have one great Linux desktop environment, instead of several okay desktop environments.
People want different things. The extra work added on different desktop environments is work that neither Microsoft or Apple is doing, and I think that's a good thing. If you want an environment for the average Joe, use desktop A. If you want a techy "fiddle all the knobs" one than use desktop B. If you want a minimalist one, use Desktop C, etc. MS doesn't make three or more versions of their OS suited for different populations, you get what they want to give you. Same for Apple.
Diversity is a strength of Linux. I don't think Pepsi worries about fragmentation with their various flavors of Pepsi, Mountain Dew, and whatever other hundreds of brands they own. Sure, diversity can be a problem if there aren't enough developers to go around but I'd rather have more options than fewer.
Same with billions of programming languages, frameworks.
Do you guys complain because there is more than 1 game studio? You'd rather have one movie/serie and a handful of actors director so all movies and series are made by the same team in order to not waste ressources? A single house architecture plan so that everyone live in the very same house with no customisation possible?
There is no one-size-fits-all solution here because there isn't a single entity controlling the development and access of Linux desktop environments. Pretty sure if all of those teams got together to somehow deliver a solution, neither the devs or the users will be happy with the result.
Use a different operating system if you want your vendor to tie you up and tell you what to do. Buy a pretty toaster if you want a slick appliance— not a computer.
I'm sick of this complaint, which boils down to this: desktop Linux (taken collectively) isn't very much like a product for consumers by a vendor who wants to suck you into their ecosystem.
Guess what? That's *what's good* about the Linux desktop.
If you want a no-choice monoculture whose design better supports a thriving hellscape of proprietary crapware, you've got options already. If you want to be a consumer instead of a participant in a community and a tradition, go buy something.
Hah, is there a middle-ground that likes choice in the Linux world without subscribing to the delusion that Mac/Windows ecosystems constitute anything a reasonable person might describe as "a thriving hellscape of proprietary crapware"?
I'll buy/build a Linux desktop when we get quality hardware and software options (where "good hardware" means more than "fast CPU! big memory!"), but for now Mac is a pretty good experience. You can't hack around the source code for proprietary programs (I've pretty much never wanted to do this on any desktop system), but you can run open source alternatives for everything except maybe some core software.
> Hah, is there a middle-ground that likes choice in the Linux world without subscribing to the delusion that Mac/Windows ecosystems constitute anything a reasonable person might describe as "a thriving hellscape of proprietary crapware"?
I realize I come off as extreme here, but that's really how I feel.
I think that's an obviously fair way to describe the Microsoft Store and the Google Play Store. (Idk about iOS' App Store because I don't use iOS.)
I'd also make the argument that it's a good way to describe the setup of basically any macOS power user, with their inevitable collection of brittle, mostly proprietary, solo dev apps they use to hack basic functionality back into Apple's anemic OS offering.
(Windows suffers from that latter problem, too, though it's not as egregious as macOS.)
> you can run open source alternatives for everything except maybe some core software.
The core software is pretty much what we're talking about here: DEs and/or bundled apps.
> for now Mac is a pretty good experience
I understand that many people like it and I think their reasons for liking it are mostly good. But for me, using macOS is genuinely miserable, not just a little off. And I think the kind of app ecosystem that Mac people really love, of thoughfully designed, hidden gems by very small teams or individual devs, distributed as proprietary software for a small fee... is just not that great. To an extent that Mac lovers rarely admit or don't understand, a huge proportion of those apps exist only to compensate for Apple's own oversights, which wouldn't matter so much in an open ecosystem. The rest just aren't worth the cost of a closed ecosystem, in my opinion.
> I'd also make the argument that it's a good way to describe the setup of basically any macOS power user, with their inevitable collection of brittle, mostly proprietary, solo dev apps they use to hack basic functionality back into Apple's anemic OS offering.
Respectfully, this seems like a really out-of-touch perspective.
The software I use for Mac is a melange of open source and proprietary. I basically just use whatever is the least painful; sometimes that's open source, sometimes it's proprietary. Virtually every piece of software I use is high quality, and Mac doesn't really encumber my ability to use best-in-class software.
Linux desktop software is by far more brittle--it has to play nicely across a dizzying array of systems all while using crumby, buggy GUI toolkits from the 90s. Basic things like hidpi, webcam, (good) touchpad support are very unlikely to work out of the box if at all. Further still, support for popular apps can be really hit-or-miss on Linux. Not sure what the state of the world is, but for a good long time Spotify wasn't available on Linux. Yeah, I get that Spotify is proprietary and thus evil, but I don't really care about ideology, I want to be able to listen to my music without jumping through hoops (I would work around the limitation by using the web interface, but it was buggier and clearly a second-class citizen).
Like, I would love to be a Linux desktop user, and I often find myself daydreaming about how to build a Linux desktop platform that doesn't suck--it's not like I'm a MacOS partisan or anything. It's just that today, MacOS is wayyyy better than the Linux desktop. Linux can get there, but we really need to get rid of a bunch of cruft and build on a saner foundation (frankly, we should just use web APIs for building apps--something like ChromeOS but without the restrictions on what a program can do).
> I'd also make the argument that it's a good way to describe the setup of basically any macOS power user, with their inevitable collection of brittle, mostly proprietary, solo dev apps they use to hack basic functionality back into Apple's anemic OS offering.
Mainly a DOS/Windows user for about five or six years, then Linux for about eight or nine (mostly Gentoo and, later, Ubuntu), macOS for my serious-business desktop needs since 2011.
To this part of your comment: Wut.
I truly have no clue what you mean by this. I think I use one program that might fit this description (Spectacle—which has several active replacements I could switch to, and still works perfectly, not so much as a single glitch, bug, or bit of jank that I've ever seen, despite its having been abandoned years ago)
Meanwhile, "brittle and relying on tons of solo-dev apps to hack in basic functionality" (ok, mostly not proprietary ones, sure) is about how I would have described desktop Linux. But... I suspect we have different definitions of "basic functionality".
Power management I never have to think about or fiddle with is table-stakes for me these days, for instance—I don't got time for that shit these days, a computer that can't do that fairly competently without my telling it what to do is just broken, same as a thermostat if I had to go poke it every single time I wanted the AC or heat to kick on, then watch carefully to make sure it actually did what I wanted, would be broken.
A good default en keyboard layout (why would the default not be a good one? It boggles the mind—and sure, that's a distro concern, not a "Linux" concern, but so's everything that matters on Linux).
"Find my" or equivalent.
Low jitter and reasonably consistent latency, at least under light load.
A trackpad good enough I don't even consider taking my mouse unless I'll be gone several days.
Bluetooth audio that works well enough that I don't hate it and spend no more than a minute or two a week fiddling with (and that mostly because I also use the same devices on Windows).
Seamless password & payment sync across all my (non-server) devices, relying on biometrics on all of them so I rarely have to type a password at all.
I'd be "hack[ing] in" all of that—and far, far more—back in on Linux, if I could attain it at all, and that stuff—the stuff that I rely on weekly, if not multiple times a day—is what I regard as "the basics".
I'd also still be using exactly one of those "thoughtfully designed, hidden gems by very small teams or individual devs, distributed as proprietary software for a small fee" text editors if I went back to Linux. Sublime beats anything else I've used on Linux, and it's not a close contest—failing that, something from Jetbrains. Open source editors and IDEs would only enter the picture if I somehow couldn't get Sublime or something from Jetbrains. Meanwhile, I struggle to think of anything else in that category that I use. It's mostly first-party or open-source. If I did more multimedia or GUI design I'd probably use a few more of those small-team proprietary programs, but I don't think it's controversial to assert that those largely blow anything available on Linux out of the water (except the ones that are cross-platform because they're—ugh—electron or browser-based, so do work on Linux) so it's not like I'm missing out on the riches of open-source, at least when it comes to that kind of thing.
> A good default en keyboard layout (why would the default not be a good one? It boggles the mind—and sure, that's a distro concern, not a "Linux" concern, but so's everything that matters on Linux).
You must have had some very bad experience with keyboard on Linux. Just curious: what was that?
Normal en keyboard layout's just not good—too hard to type various characters that occur in English text, some uncommon-but-not-rare and some downright common (—, ü, å, ¢, °, •, é, ç, ñ, and so on, and sure, a bunch of those are for loan words, but that's still English).
Linux has good ones, I've just never seen one as the default; if you just click through the "happy path" you'll have a crap one, and there's little guidance on which you should pick if you need to, like, actually compose text in your native human language, so the user just has to know in advance or go look up which one to select.
Mac's default, meanwhile, is (at least) nearly as good as the best Linux has, so you can start fluently typing English without having to fiddle with settings or memorize weird number sequences (sure, there's still some memorization, but much of it's simpler and closer to making sense and far more guess-able than, say, a four-digit string of numbers). Last I checked, Windows gets this wrong, too. Why either of those operating systems does that, I have no idea. That it was even incorrect didn't occur to me, somehow, until I switched to Mac and was like "why isn't the default at least this good on every platform!?"
Linux has the "Compose key" concept that solves 99% of what you list in your claim. You have to map it to some physical key on PC keyboards, though. All the DEs I checked let you do this in the settings. My only complaint is they they don't allow out-of-the-box to assign it to the Insert key, which I never use in its intended function.
I actually agree with you, but I think that's not what people complain about. I think people mostly complain about a lack of cohesion that Windows/macOS have.
But then again, their design language has recently started to be less cohesive too (or more precisely, mix of older and newer design systems are apparent).
Like, it's generally difficult to get something working/looking nicely and consistently on Linux. A somewhat exception to this rule is GNOME, but even then there's issues with KDE/Qt app theming sometimes.
Like, until 2021 (I think) I lived with a broken looking inkscape and Krita (used it as a hobby) because I gave up trying to figure out how to fix the theming, since none of the fixes suggested online worked. Then one day after some random update it started working
So, there's a lot of space for improvement and better cross compatibility between these "subsystems"
Granted, though, it's getting better. But I assume not at a pace an average user can perceive. So, i thinking that's where people get frustrated.
The Windows ecosystem has never had a coherent design. I fondly remember the Windows 7 days where we had Windows Media Player, Windows Media Center, and Zune Media Player each with a completely different design style despite coming from the same company. Similarly, Outlook vs Windows Live Mail vs etc or Word Perfect vs Word. And these are just comparing similar Microsoft products.
Ok I guess I've had a bit of a different perception back then, since thinking back I think you're rights.
I guess the thing I was thinking of being cohesive were stuff like context menus, or integrating better with the window manager (or whatever it's called on Windows) than they do on Linux (since you had to integrate with the only existing one, not choose one). My earlier complaint about Krita/Inkscape is that most "labels" rendered in a colour and labels that was almost indistinguishable from the background, so I was largely working with shortcuts, or from memory and sometimes squinting to see what I'm clicking lol. It was an issue with all non-gnome apps, but I mostly used those two.
Back in the days even on windows you had java apps with different toolkit / experience. Now you have electron apps, there isn't even much cohesion.
I quitr remember that on Mac it is barely better.
Now take some specialized areas such as DAW, video editing software and 3D editors and you can just throw that idea of cohesion out of the window regardless of the OS.
Yep, I think cohesion is only going to die going forward, and I think web technologies are going to be what kills it. Electron is already popular, and nowadays every platform has a system web browser that can run in a "webview" mode so you don't have to bundle a full browser with each app. And building apps with HTML/CSS/JS is wayyyyyy easier than using native toolkits or cross-platform toolkits, especially if you want to target multiple platforms.
Hmm. I think that devs developing for macs tend to at least try and reuse established design language forms. I've only recently started using a mac for work and a lot of stuff I use seems to try and feel "maccy" (with varying degrees of success). But then again, I don't use many mac apps either, so I could have a poor sample rate and was recommended good looking apps by colleagues
On Linux, if you used Gnome but ran a KDE app (or were on Gnome but ran a GTK app) you'd by default get something that looked horrendous broken visually but worked, and sometimes the visuals actually managed to break the app (ie thing being out of proportion, or invisible etc). It would look great and cohesive if you ran a GTK app on Gnome, or a KDE app on KDE, etc.
I think it's gotten much better at some point when these environments decided to support each others themes, though.
I wish they all sat down and came up with a unified theming framework for apps (maybe based on CSS, since they kinda use CSS flavours already), but have the rest be implementation details. Dunno how realistic that is, though.
Windows I think I missed the mark actually. A lot of people pointed out stuff I didn't think about or forgot about (every Office Suite release looked different, Java apps etc, branded stuff like Adobe or specialist software, etc).
I think it might be because I never really used those apps that my much, since Windows was basically only my gaming OS
> But then again, their design language has recently started to be less cohesive too (or more precisely, mix of older and newer design systems are apparent).
It's the age of Electron, baby! For better (more frequent cross-platform support, rapid app development, empowering the masses of web devs to create desktop apps without retooling) and for worse (non-native look-and-feel, high resource consumption). Nobody really has a cohesive collection of apps on their desktops anymore, not even Mac people.
> I think people mostly complain about a lack of cohesion that Windows/macOS have.
Like we've both noted to some extent here, I think this is pretty much a thing of the past. It can be pretty good on Linux, at the same time. If you stick to KDE or Gnome (which I think are both reasonable propositions), your desktop on Linux will be way, way more cohesive than Windows has ever been in my lifetime (maybe ever).
In a strange turn of events I think you might have a better shot at a cohesive experience on a carefully curated Linux desktop these days than macOS. Even Apple's first party apps are now a mix mash of different UI styles. Once you start installing third party apps you get something that looks more like a Linux desktop from 15 years ago than the glory days of OSX.
In fairness I think I do know what they want, which is better support from third-party (and especially proprietary) software vendors. A Linux workstation can be painfully close to perfectly usable for a lot of professionals who have various personal reasons to prefer it to Windows or macOS— reasons you and I would likely emphatically agree with, no less!
There's this hope that if only for the fragmentation, Linux might finally have Photoshop or whatever.
I get it. I've contended with integrating disparate GUI frameworks on my system. I appreciate what Apple's virtual monopoly on app frameworks for macOS allows them to do in terms of accessibility and integrations and UI changes, in rapid, uniform ways.
But you can't impose uniformity on the Linux desktop without draining the oasis, without killing what makes it a breath of fresh air in the first place.
I don't think it's even the UI that's the problem.
The problem is this:
> ./GuitarPro
./GuitarPro: error while loading shared libraries: libssl.so.0.9.8: cannot open shared object file: No such file or directory
I bought this program several years ago while using a Mint 13 system - it won't run on Mint 20. Maybe I could set up a Docker, or copy the correct version of the libraries, or whatever - maybe it could be coaxed into working, maybe not - but while end users are likely to encounter this kind of roadblock Linux is just a non-starter as a platform for commercial software.
> There's this hope that if only for the fragmentation, Linux might finally have Photoshop or whatever.
If Adobe builds Photoshop for Red Hat and make it run only on Red Hat, people wanting to use Photoshop will install Red Hat. The big problem is what happens when Microsoft makes Excel work perfectly on Ubuntu and only on Ubuntu.
> The big problem is what happens when Microsoft makes Excel work perfectly on Ubuntu and only on Ubuntu.
This is basically how Steam still works, pending the release of SteamOS 3.0, and it hasn't been a problem for other distros. They can sub in their own libs and repackage the thing or use an Ubuntu chroot.
This is what Snap and Flatpak are for though, and I think they'll handle it well, going forward.
