I'm glad that System76 wants to create their own DE, with actual graphics/UX designers.
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.
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.
So System76 is going to go NIH. Let's see how that goes.
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.
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.
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. :)
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.
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.
Pop_OS has added a lot of this crucially missing UX. If they can build a better mouse trap, I wholeheartedly welcome it.
What do you mean?
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.
https://boilingsteam.com/which-linux-distro-for-gaming-q2-20...
those distos have undeniably won the server market.
But they also don't really care about the desktop market as a result.
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.
It is absolutely not an interesting distro in itself, from a technical point of view.
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.
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.
Tried to use it anyway, but it's unmaintained and instable now.
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.
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.
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?)
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.
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.
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.
I want fast gnome 3 without the crashes, and consistent modern ui themes across my apps.
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.