As an exercise, I threw OpenBSD and GNUStep on a machine the other day. I wouldn’t say GNUStep provides the best Free/Open desktop, but I want to believe that it could.
The thing is, many of the modern Linux desktops seem to be heavily wedded to a lot of dconf and systemd infrastructure. OpenBSD has managed to build analogs for enough of the heavily Linux-centric infrastructure that Gnome 3 depends on, but man it’s clear that Gnome 3 was written with Linux and only Linux in mind.
I didn’t see KDE in the OpenBSD ports when I looked, and I suspect it too is because KDE has become wedded to Linux-only services.
XFCE remains the most portable full-featured Desktop environment and runs well on OpenBSD.
But GNUStep... GNUStep is my biggest regret. GNUStep isn’t entirely comparable to XFCE, because GNUStep isn’t a desktop. It’s more like all GTK plus dconf plus all the desktop services required for interaction between GUI apps plus a uniform display layer.
In short, GNUStep is more or less like Cocoa from macOS. Add GWorkspace, and you have a solid reimplementation of macOS’s Finder.
I wish GNUStep had caught on or that someone would inject new life into it. There are apps that compile on both GNUstep and macOS. For whatever reason macOS seems to attract better desktop apps—both Free and commercial.
Had GNUStep and not KDE or Gnome become the defacto *Nix desktop, I imagine people would be writing macOS apps for free on Linux, and macOS users would be recompiling apps to run on Linux... and not just Linux, but all the platforms that GNUStep supports, like OpenBSD and Windows.
EDIT: Oh yeah, and why I thought to write this, is because GTK vs QT both feel like lower potential toolkits. And GNUStep also feels very lightweight and portable.
> I didn’t see KDE in the OpenBSD ports when I looked, and I suspect it too is because KDE has become wedded to Linux-only services.
I doubt this is the issue. KDE Plasma 5 seemed okay on FreeBSD when I tried it recently, although I don't have any recent KDE+Linux experience to compare it with. More likely there just hasn't been enough developer interest to get it properly ported to OpenBSD.
Edit: To get more at your actual point, I've had similar feelings about GNUStep, but when I dug into it I felt like there was just too much of a mismatch with the rest of the FOSS GUI ecosystem. Maybe I'd be a die-hard devotee if I'd tried it back in the '90s as my first thing after fvwm-configured-to-look-like-mwm.
> I felt like there was just too much of a mismatch with the rest of the FOSS GUI ecosystem.
Definitely. It probably has the largest mismatch outside of exotic options like, I don’t know, a Smalltalk image.
I think that’s why the Free *nix desktop struggles to this day. We never built the core infrastructure that allowed useful inter-application messaging. We never built the application frameworks to enable a really great, consistent UX. When we finally did, we built dconf and systemd.
I can’t say these things are horrible, but neither can I say they offer anything more than macOS’s system services or Cocoa APIs. And the price of being different, is that macOS is far and away the best GUI platform today (I’m not a fan of the Mac these days for other reasons, but the UI consistency is first rate, owing to their frameworks and the NeXT legacy). We could have shared (maybe still could share) so much with the Mac. And importantly, we could have benefited from Apple’s singular vision while also providing an escape from the walled garden.
But yeah, the FOSS GUI ecosystem chased FVWM bling for a long time. Oh my goodness, I spent hours tweaking FVWM configs, later Window Maker, Enlightenment, and who knows what else. But it was all skin-deep looks. We spent so long chasing window themes, but ignoring the hard work of building infrastructure.
KDE and Gnome were ultimately late, divisive, and different from everything else. So much of Linux and the BSDs is great, it bums me to see us so far behind Apple, when we could have chased tail lights until we overtook them.
I remember that I was a kid when I read the news on Slashdot that the Linux desktop experience was going to be basically a competition between Gnome and KDE.
At that moment I knew that Linux would never win on the desktop. All those wasted developer hours.
I am still heartbroken to the day. In capitalism, competition benefits the user, even if the products are virtually identical, because then they compete on price. In the free software world, competition of two indistinguishable products actively harms the user.
The thing is, however, that GNOME and KDE are not indistinguishable - they each have very different goals, ideas on how the desktop should be, how to design applications and UI, what sort of technologies to use and develop, etc and all these end up making two very different desktop environments. As a result, they (and other DEs and WMs, of course) cater to different users and while there are some users that can use both (and all, or at least most of the other DEs/WMs) this sort of "competition" helps the users who align more with what each DE (and WM) provides. In other words, they provide options.
There are people who prefer GNOME over KDE, would you force those people to use KDE? There are people who prefer KDE over GNOME, would you force those people to use GNOME? There are people who prefer Window Maker over any desktop environment, would you force them to use KDE or GNOME? There are people who prefer i3 over any DE or WM that uses an overlapping window UI paradigm, would you force them to use KDE or GNOME or Window Maker?
If you did so, you'd be making their experience with their computers worse - you'd be acting against their choices. If people didn't value having options, they'd flock to a single desktop environment (or WM) and the rest would be a mere curiosity at most. But this doesn't happen.
And of course what Slashdot predicted didn't happen either: in addition to GNOME and KDE we also have MATE, XFCE, Budgie, Pantheon, Enlightenment, Cinnamon, LXDE/LXQt and a TON of standalone window managers to choose from - we even got the rising popularity of tiled window managers, which was certainly not a thing back when KDE and GNOME were new.
Of course, they have different goals, they're using different frameworks and so on. But ask yourself this. If the average Windows user had a choice of two different desktop environments, would they be more likely to be thankful for having a choice, or is it more likely that they would be confused?
Windows and Mac OS X don't allow much customization. Not because Microsoft and Apple are lazy, but because they want the average user to be able to pick up any Windows PC or Mac and be immediately familiar with how to use it.
If we didn't have the Gnome and KDE wars, we would have likely ended with a "standard" Linux desktop.
Considering the popularity of tools like Classic Shell, how much people despite major changes to the Windows UI and even to the lengths some people went to make the classic theme usable on Windows 8 and 10... yes, i'm 100% certain that there would be a lot of users thankful for having a choice.
There is this guy making a DE he calls NEXTSPACE[1] by combining Window Maker, GNUstep and a bunch of custom applications and modifications to existing stuff for them to work better together. I haven't tried it, but i've seen it be mentioned a few times the last couple of years.
I really miss GNUStep. Pretty much the only time I remember actually kinda enjoying my Linux Desktop experience was when I had some ancient (found in a dumpster) laptop I set up with debian and WindowMaker+GNUStep. Granted, I didn't use it for much, but those guys at least had a lot of the right ideas going that all the other Linux Desktop Environments continue to ignore and pretend no one cares about.
Imagining an alternative timeline where GNUStep became the dominant Linux Desktop makes me feel like I live in the darkest timeline.
Honestly I remember trying gnustep, it was ugly. I can't even emphasis how terrible it looked from a ux/UI perspective compared to literally every other desktop.
Yup, they were doing the hard part of making application frameworks and consistent APIs, and were always deferring window dressing. By contrast, old school FVWM and later Enlightenment were pushing limits of themability without really building anything of substance beyond literal window dressing.
No reason GNUStep couldn’t be as beautiful as anything else. Oddly, I find the NeXT aesthetic oddly fresh looking today. I was struck by how good it looked when I took a trip to the Living Computer Museum and played on their NeXT cube. What’s old is new I guess.
Still, not supporting themes from the getgo was definitely a reason GNUStep didn’t get traction early on, I’d
I stood in one FOSDEM presentation about 10 years ago, where they spent a couple of hours speaking and demoing part of it, but it never seemed to have moved beyond those presentations.
Next step in my experience was crashing constantly and ugly. It was the basis for macos. Which permanently turned me away from apple products in the 90s. Rebooting a frozen computer should not require literally pulling the plug.
I wouldn't say they don't want to modernise it. They have a theming engine that can use some other frameworks' GUIs to provide theming (like GTK+, Windows, and probably Cocoa) — but most distributions only have a hideously ancient version of GNUstep that lacks the theme engine.
I'm with you on the general idea that the GNUstep developers don't seem to be in a rush to take GNUstep out of the shadows and let it be a glory on Linux. More aggressively imitating Cocoa, adopting Swift and filling in some of its holes on Linux, and abandoning the wholly useless enterprise of source-level compatibility with the OPENSTEP APIs of the 1990s… these would be good.
The thing is, many of the modern Linux desktops seem to be heavily wedded to a lot of dconf and systemd infrastructure. OpenBSD has managed to build analogs for enough of the heavily Linux-centric infrastructure that Gnome 3 depends on, but man it’s clear that Gnome 3 was written with Linux and only Linux in mind.
I didn’t see KDE in the OpenBSD ports when I looked, and I suspect it too is because KDE has become wedded to Linux-only services.
XFCE remains the most portable full-featured Desktop environment and runs well on OpenBSD.
But GNUStep... GNUStep is my biggest regret. GNUStep isn’t entirely comparable to XFCE, because GNUStep isn’t a desktop. It’s more like all GTK plus dconf plus all the desktop services required for interaction between GUI apps plus a uniform display layer.
In short, GNUStep is more or less like Cocoa from macOS. Add GWorkspace, and you have a solid reimplementation of macOS’s Finder.
I wish GNUStep had caught on or that someone would inject new life into it. There are apps that compile on both GNUstep and macOS. For whatever reason macOS seems to attract better desktop apps—both Free and commercial.
Had GNUStep and not KDE or Gnome become the defacto *Nix desktop, I imagine people would be writing macOS apps for free on Linux, and macOS users would be recompiling apps to run on Linux... and not just Linux, but all the platforms that GNUStep supports, like OpenBSD and Windows.
EDIT: Oh yeah, and why I thought to write this, is because GTK vs QT both feel like lower potential toolkits. And GNUStep also feels very lightweight and portable.