I have never played with GNUStep. By the time I actually started real work as a professional software person (2011) it was already kind of considered a joke, so I never bothered learning how to use it.
It bothers me a bit, though. Developing for desktop Linux is still a pain in the ass, and I really wish the Linux community had agreed on One Desktop Framework To Rule Them All, and I think GNUStep could have been that framework if the community had been willing to embrace it.
> I have never played with GNUStep. By the time I actually started real work as a professional software person (2011) it was already kind of considered a joke, so I never bothered learning how to use it.
I dunno if it was a joke or not, but I can confirm that I used Windowmaker as my daily driver from around 2004-ish to 2019-ish.[1]
The only reason I switched, and did not switch back, is due wanting my virtual desktops to be arranged in a 3x3 grid, and Windowmaker did not have that, nor was it on any roadmap.
The thing I miss about my Windowmaker days is how focused it let me be - the entire screen was devoted to a window (or multiple windows); nothing crying out for attention in some bar at the bottom of the screen.
Should I get a few free moments in the near future, I'm tempted to actually add that feature in.
All the other things (taskbar, wifi/sound/etc controls in the taskbar) I can either do without (i.e. I don't need it if I have a decent launcher that is bound to a shortcut), or I can use an existing dockapp.
=======================================
[1] Here's a video: https://www.youtube.com/watch?v=_W_4fH_ccQE
You might like something like i3 (if using X) or Sway (if using Wayland), or another tiling manager like Xmonad.
They are extremely minimal and really get out of your way. I had to write my own service to power the info on the Sway bar [1] because the entire thing is so minimal.
For me, I have title bars in windows disabled, and the gap between different windows is literally 1 pixel.
[1] First version: https://github.com/tombert/swanbar Rust rewrite: https://github.com/tombert/rs-swanbar
WindowMaker certainly was quite popular for a while, probably also because it can be used without GNUstep, just as a plain X11 window manager.
Actually, it has its own toolkit named WINGs. Which stands for WINGs is not GNUstep ;)
https://www.windowmaker.org/docs/wings.html
> Developing for desktop Linux is still a pain in the ass, and I really wish the Linux community had agreed on One Desktop Framework To Rule Them All
Not going to happen. And even if it would happen, very soon there would be a split, because some super important addition to CoC wasn't accepted, or because someone committed a wrongthink on a public platform, or because some patch of snowflake dev superstar was rejected, or whatever else.
Hell, even Microsoft cannot agree with Microsoft on One Desktop Framework To Rule Them All, so it is hard to expect that from the Linux crowd.
At least with Microsoft there is still Win32, with Linux fragmentation not even that, Xlib and Athena are long gone, and aren't full stack as Win32.
I think that most active, current Linux users see fragmentation as a good thing, because they absolutely don't want consolidation of power. Many will point to a time when GNOME reigned supreme, and how did they use that power? To repeatedly completely redesign the GUI in ways that the community hated, because "we know better".
There is a much larger group of people who would support "One Desktop Framework To Rule Them All", but they aren't as loud, so they don't get heard.
Wait, did Gnome ever reign supreme? I feel like there was always a back and forth with Gnome and KDE?
I'd say that the momentum in the 1999-2001 timeframe was almost overwhelming, with a lot of buzz around the likes of Eazel and Helixcode, together with the excitement of Sun choosing it for Solaris.
It was certainly enough to push GNUstep/Windowmaker and Enlightenment into "also-ran" territory.
If Gnome 2 had come out a year before the dotcom crash rather than the year after, I think it probably would have definitively overtaken KDE as well. In the event, it didn't work out that way - and, as you say, KDE did enough to hang on (thanks largely to Konqueror!)
Meanwhile many of the folks responsible for those companies are now working with Microsoft and Apple technologies, which kind of tells how much they runned out of steam trying to make a useful GNU/Linux desktop.
KDE was going equally strong at the time though. Anecdotally, it feels like more people were running KDE 2/3 around that time period than Gnome. And while Gnome did get an overall user count boost from becoming the "enterprise Linux DE", that same thing also turned off some people, with projects like Xfce getting a boost. So I wouldn't say it ever really "reign supreme".
Never fully, but it's how we got systemd nearly everywhere - GNOME 3.8 arrived with support for polkit-based approach for, among other things, power management, and mandated logind. Which at the time was very much a banana that required the gorilla and the whole jungle to run. Some tried to deal with it, some managed to even separate udev and logind out, but the immediate effect was that distributions that avoided systemd or made it optional - including ones that were exploring alternative paths, were forced to adopt systemd or remove support for GNOME 3.8 and newer.
The founder of the GNOME project founded Mono to be the “One Desktop Framework To Rule Them All”. It did not work out.
That was to be expected, there was a lot of resistance against using C# because Microsoft tried to destroy Linux in various ways. Toolkit-wise, the Mono on Linux stack used Gtk+. So given the C#-distrust it was easier for people to use C, Perl, Python, or later Vala.
But what if that One Desktop Framework To Rule Them All had sucked? :)
Definitely wouldn't be worse than multiple desktop frameworks that suck.
Current desktop frameworks are quite good. KDE/Qt are incredibly mature at this point. GTK still has a decent amount of churn.
It definitely would be.
Choice is a defining factor in freedom.
> Developing for desktop Linux is still a pain in the ass
GNOME works great if your language has good bindings for it. Qt is a bit more iffy but still usable if you need cross-platform.
I would argue that it’s heavily in the other direction, having developed fairly extensively in both.
The QT Linux ecosystem is far more cohesive and consistent and QT apps work more seamlessly between KDE, Windows, and MacOS. In my opinion, at least.
Agreed. Qt, for all it's flaws, focused on cross platform. Despite getting sold/resold nearly to oblivion, it had a commercial end (some paying customers), and so it received a level of polish that gtk just never did.
Not to mention, when the gtk3 devs went off the deepend, completely broke backwards compat so they could try some new UI...and you couldn't consistently run gtk on Linux...
I wish there was a good Rust binding for classic Qt...
It only having first class bindings for C++ and Python is a problem for a lot of folks, though. Plenty don’t want to write either.