All existing DEs suck, partly because programmers don't know how to design UI/UX, partly because the one project with actual graphics designers (GNOME) wants to do something completely different than what most end users want.

I'm glad that System76 wants to create their own DE, with actual graphics/UX designers.

I think GNOME actually gets a lot right, perhaps even more than KDE for the average use case, but has a few high visibility misses and perhaps suffers from muddled focus by trying to also accommodate smartphones.

With a few key changes like bringing back the minimize button by default, integrating the dash-to-dock extension by default, and dropping the crusade against proper menubars it’d be easily the most well rounded and DE with the most usable out-of-the-box experience.

KDE shares the problem of muddled focus between desktop and mobile, but instead of lacking functionality instead has issues with its design feeling somewhat amateur, as well as bad defaults. KDE can be great if tuned extensively but without tuning it’s a mess.

I do agree that’s it’s good to have more players in the DE space, because no existing option has fully cracked it and more options is always good. It would be nice if more DEs shared work between each other (is it really necessary for there to be 3+ forks of Nautilus?) but oh well.

What I want to see most is a new UI library. Qt is nice from a technical perspective, but it being tied to C++ and thus lacking quality, up to date, idiomatic bindings for other languages really hurts it, as does its licensing shuffle. GTK is wonderfully flexible because it has bindings for almost every language ever, but comes with frequent breaking changes and forced GNOMEisms, and feels awkward outside of Linux. The ideal UI library would have GTK’s flexibility with QT’s technical strengths and platform adaptivity.

I always had dash to dock and all the other small common tweaks that everyone makes.

Until about a half year ago, when I decided to stay vanilla on a fresh install. After a week, I forgot why I needed all these tweaks.

I think the far majority people, myself included, can't even tell the difference in UX quality (as long as it is above a certain level) and will mostly like or dislike based on what they are used to.

I think you hit it spot on.

Personally I think if KDE had half the resources that were put into Gnome by various distros over the years, Linux desktop would be a solved problem.

So System76 is going to go NIH. Let's see how that goes.

> All existing DEs suck

I feel almost the opposite. I think Plasma and GNOME are both really good, and some others I haven't bothered to try yet because I'm pretty happy on Plasma (Budgie, Pantheon) seem thoughtfully designed and reasonably featureful. Plus there are good tiling WMs you can use on any of them!

I feel like there are more good options on Linux than anywhere else. macOS has designers galore and it falls flat on core window management functionality, like its stupid minimize/maximize behavior, lack of window snapping, lack of tiling, lack of click-anywhere resizing, and other DE basics, like configuring global keyboard shortcuts.

But I welcome System76's effort, too! I hope it's so good it makes me switch. :)

>partly because the one project with actual graphics designers (GNOME) wants to do something completely different than what most end users want.

We should learn our lessons and don't trust some designer that probably does not use the project to design good UX, a designer should use the project(app or webpage or DE heavily) before deciding ro remove some feature just because he read in a book that too much options can cause issue to some mythical average user.

It would still be less work to build on that foundations. Heck, System76 couldn't even bother to pony up the theming specification they were supposed to do for KDE and Gnome. Never happened.

If they don't even have the resources to work on the toolkit level, how do they expect to get the resources to work on a full desktop stack?

Bad call, imho. But hey - I think Rust is an excellent choice.

Their approach seems to be piecemeal, so I don't think there is a risk of empty promises here. My guess (based on following their dev comments for a few months) is that they are just going to replace the components most diverging from their desktop vision, until it's a new desktop environment.

I think GNOME's detractors are really vocal but the vast, vast majority of users are happy, or they'd be using KDE.

There is a lot of truth in this. If gnome really was unusable and critically broken, why is it the default on almost every distro.

GNOME is not the only criterion people use to choose a distribution. I choose my distributions based on package managers, hardware support, and stability. For package managers I only feel good with dpkg or pacman, which automatically eliminates a lot of other distros. Not all of the remainder has his hardware support, and out of that, not all of those have good stability (with updates, or software bugs). This means that for me, all that remains on the playing field are Arch Linux (because I'm more willing to tolerate it), and Ubuntu derivatives.

If the price of having to use e.g. Ubuntu is having to run Gnome, I'll pay it just to get the other benefits.

Most distros provide DE versions and yet the gnome one is more popular with all other factors the same.

Kubuntu is an official flavour of Ubuntu; you're not locked in by any means.

> All existing DEs suck, partly because programmers don't know how to design UI/UX, partly because the one project with actual graphics designers (GNOME) wants to do something completely different than what most end users want.

I guess that here we really need to figure out which end users we are talking about, to understand what they actually want.

To me, and perhaps some others, DEs like XFCE are pretty close to perfection - low resource usage, good performance, good usability, decent customization (as long as you don't want to do anything fancy, just get your windows to show up in a panel or two, and add a few widgets for resource display, clock, language switching etc.).

To users like me, the finer points of the DEs out there that have more fancy graphical elements or in depth customization, like KDE, are pretty much lost or are not really relevant. In my experience, most of the problems with the modern Linux desktop experience are created by the software, that has inconsistent element placements, varying button sizes, different UI frameworks altogether and so on. However, as long as the software works, i also don't find myself caring too much, even if things could be better.

I'm gonna go down in the flames here but Unity was great. It was very stable for one thing. I never had an issue that I do with literally every single other DE that exists. XFCE is stable but customization is a time consuming affair. KDE is riddled with bugs. Cinnamon doesn't get enough support and thus has issues with various software (recently like not being able to run Factorio via steam or having bloated font in KeepassXC), and GNOME is like a happy medium between KDE and XFCE. Unity just worked. Out of the box like Windows and MacOS. It made me sad cannonical decided to drop it. It was the first real dedicated support to a DE that Linux decidedly needed.

I agree, I loved unity and miss it.

Especially, there global menu and HUD support.

nah, Unity was great

I don't totally agree. This can work. Gnome and kde have a larger mission. They need to be everything for everyone. They need to work on many different archs. You end up either special casing everything, or being the lowest common denominator. Pop_os is carving out its own niche. System76 can just focus on what its customers want. And its customers are a much smaller subset from all of the users/systems running kde/gnome.

Wholeheartedly agree. Gnome in particular I have been very disappointed/baffled by in the past few years. Killing off desktop icons, system tray/status icons, pushing out global menu bars. Not to mention the way they handle HiDPI and fractional scaling is atrocious.

Pop_OS has added a lot of this crucially missing UX. If they can build a better mouse trap, I wholeheartedly welcome it.

> Not to mention the way they handle HiDPI and fractional scaling is atrocious.

What do you mean?

I'd expect System76 customers to use their machines for the same range of tasks as most other desktop Linux users. Perhaps they don't need to support as wide a range of hardware, but how much does that matter for a DE?

There's an open space to fill right now for "switchers" coming from Mac OS, with a high expectation of polished experience when they invest more than $2k into their main driver hardware. These new Linux users don't want "choice at all costs" (the current vibe I get from the community) but rather, "good design that just works", with a Mac OS expectation origin. I think this explains why System76 chose Gnome over KDE, and why they built Cosmic. If their users aren't satisfied with Gnome's (somewhat inflexible) constraints, then the natural choice is to piecemeal develop a better DE for customers.

I remember when PopOS was first announced and everyone was like "terrible decision, just use one of the many distros already out there." Yet, PopOS is now one of the most popular and most recommended distros out there.

Without downplaying the work put on Pop_OS, anyone can start an Ubuntu derivative distro in a few hours. Writing a DE from scratch is something else entirely, Ubuntu famously tried and it went nowhere.

It is a completely different undertaking but my point is, I'm willing to give System76 credit here and wait and see how this plays out instead of jumping to cynical conclusions about how terrible this will be.

>Yet, PopOS is now one of the most popular and most recommended distros out there.

among enthusiasts on the internet possibly, but in terms of real world usage it is ubuntu/debian/centos etc.. followed by a long list of nothing.

there's always a new distro for tinkerers but five years later you never hear of it again, and when you invest into building an entire Desktop Environment that's supposed to compete with Gnome or KDE you're probably looking at that timescale before it's even competitive.

It's a gaming survey with mostly reddit users, so may be skewed.

https://boilingsteam.com/which-linux-distro-for-gaming-q2-20...

ubu/debian/centos: do you mean desktop, or are you adding in servers?

those distos have undeniably won the server market.

But they also don't really care about the desktop market as a result.

It seems Canonical can make even worse decisions - for me it was going all in on snap

Is it helping their business though?

Pop OS is the best Linux desktop right now for new users. Imagine if they would have decided to not built it at all. Quite a loss for Linux users in general.

And now they want to build something new from scratch... The possibilities are pretty huge. In a few years we could have something that is faster and better looking than mac os for Linux, with innovative new features that doesnt exist in mac os or Windows.

Also great to see this happen from System76 and not Google or Microsoft. The culture of a company is super important for what it produces.

I mean, Pop OS is just ubuntu with a few tweaks here and there — not taking away from them but let’s be honest.

It is absolutely not an interesting distro in itself, from a technical point of view.

Nobody should try anything new in the DE space because more mature things have been established with far more engineering hours?

That's odd to me. You don't think they'll look at the design and learn from the advances made during those centuries of hours? That's hardly throwing it all away.

Nothing has been "established" in the linux desktop. It's comprehensively lost in the marketplace for the last two decades. Gnome and KDE are at this point losing approaches, both have scrubbed progress several times for major rewrite revisions that promised several steps forward for one step back. Problem is, they never really made the steps forward in terms of market share.

Linux can't even win the software developer wars. How is OSX, which can barely run docker, has a linux incompatible CLI, and requires relearning keybindings if you come from windows or linux desktops, and is steadily jailing its applications into a store, winning the desktop OS wars among software devs that are probably all running linux derivatives in K8s or VMs?

A good team that knows DEs and the current DE codebases can probably work wonders.

It has been more than 10 years since Maverick Meerkat. About time to repeat the mistakes of the past.

I disagree, I loved unity. I miss it a lot while using gnome shell.

Tried to use it anyway, but it's unmaintained and instable now.

I think unity matured nicely, but natty in particular was just about unusable.

Yep - Linux folks like repeating the same mistakes over and over.

It's part of establishing a brand for themselves. They want to be the face of the modern linux desktop. They've done a good job of this already with the tweaks they've made to GNOME (I run Pop_OS and I have had a very positive experience it), but if they really want to establish themselves, they need to create a branded experience all their own, à la ElementaryOS. A custom DE would go a long way towards providing that. It's also an opportunity to address and fix a lot of the UX mistakes other DEs have made, which is likely to generate lots of buzz (as evidenced by this thread).

Totally agree. I have a system 76 laptop and I find it confusing that they will spend their development cycles this way.

Their business model is rebadging hardware and ensuring it will run Linux well. Their value add is in working on proper hardware support for these devices in Linux and in replacing the proprietary Clevo BIOS.

I'll gladly pay more for that!

But Pop_OS in general (and now this DE) seem to be straying far afield. I will never use either. To the extent that these endeavors raise their unit prices, they risk pushing customers to directly buy a Clevo or another rebadge.

This can be said about anything. For example, using Linux.

Windows has centuries of engineering hours poured into it. It works and is reliable. There's no reason to throw it away and use Linux. If you think Linux is free, wait until you see how much you pay per year in salaries for people whose whole job is to manage this OS.

Also, all existing DEs suck in some way or another. KDE is too bloated and slow. XFCE is too minimal. Gnome has countless bugs, and its performance is questionable.

You could say this about anything that has been around right? Basically an argument to say "Don't innovate, X has been around for a long time and works!"

Even though I'd like to see Rust everything I agree, considering the amount of issues even the major DE have I find it very unlikely that system76 would be able to do any better. Improving the current DEs would be much smarter not just in terms of compatibility

eh you don't seem to realize how shitty current (GTK/Qt) toolkits are from developer ergonomics.

But you can’t really not use these toolkits — they do an astonishing amount of work that you simply can’t replace without years of work.

The actual graphical part is easy, as well as using an existing toolkit. Writing a new one from scratch is ridiculously hard (but I guess they can reuse some libs?)

I am not a dev, but talking with them and using tons of apps I'd say I haven't seen a good desktop GUI toolkit from anyone yet

Yes, but this is in Rust.

More seriously, it's an interesting decision. The linked thread has no information about why they would have made such a choice, and I doubt they made this decision lightly.

I'm curious to see how they tackle this problem.

its really not that hard. existing stuff make this a fairly easily solvable problem.

all you need for a DE is basically the wayland server, an application launcher, system tray, and settings programs.

I've been building one myself over the last couple months on top of sway (so I don't even need the wayland server). and its day to day useable for myself. two/three solid dedicated engineers can build one fairly quickly (2 - 3years).

hardest part is the UI toolkit, and there are some reasonable choices in rust. they're young but decent bases to get started.

In addition, based on the last time I looked at System76's Github, they have a lot of in house Rust talent already.

This is including the guy doing Redox OS - which means there's even some experience writing a DE in Rust, though not a Linux DE.

Aren’t most of the rust toolkits immediate mode ones? It is probably not the best idea for a DE that has to run 0-24, as that would kill the battery very very fast.

They all crash, eat performance, or ugly.

I want fast gnome 3 without the crashes, and consistent modern ui themes across my apps.

Gnome 41 hasn’t crashed on me in a very long time. Was it perhaps something bad with an extension? (Though I believe it doesn’t crash the whole thing usually)

Maybe we'll finally get thumbnails in the file picker though.

Why? There's Ubuntu spins of every DE, what's currently unique about Pop_OS? Sometimes being different is better than actually being better.

Why does Pop_OS even exist?

it's a bold move. yes. from business perspective, a new DE would not bring much benefits unless it's hugely successful.

They need maintaining this to somehow be less work than maintaining GNOME Shell extensions

If nothing else it’s good advertising.

> Existing DEs have centuries of collective engineering hours poured into them.

Some of hem have even thrown a lot of that collective engineering hours away by deciding to rewrite entire parts!

> They work.

Arguably, but I say they sometimes don't. That alone is worth considering alternatives.

> There is no compelling reason to throw that all away.

Sure there is. This one is written in rust and the others support javashit.

> Likely a new system will be worse along every metric of concern for the foreseeable future

Ahh you must have short sightedness. It won't be hard to improve performance of a few key metrics:

* reduced crashiness * reduced memory footprint * improved UI responsiveness * cold boot to desktop time

It shouldn't be that hard to be an improvement over Gnome in memory utilization or KDE in crashiness.