The point, if I am reading it right, is to create something so good, work so well, and solves the donation/purchase/monetization issues that it could plausibly be adapted to be THE default. Become THE Flathub GUI used by everyone, solving the “how do I get my apps?” issue and having something on par with the other app store apps on other OS.
At the same time, what’s also going on is that we don’t want to push it THAT hard yet. Incubate it in Bluefin, but we don’t want any rhetorics that could jeopardize the relationship necessary to make such a thing possible.
I personally liked the idea of NixOS, but it is just too annoying troubleshooting everything. I like the system of building your image on GitHub, with a yaml-style (or Containerfile-style, if that’s your thing) file.
First, it’s just more understandable - even with the Containerfile format, it is a universal format used by the industry, so it has more resources (whereas in my experience, even NixOS diehards admits there are issues with documentation and learning the format). Next, if there’s any error, you’re not stuck with a broken config file on your machine - everything is natively on git/github so it is easier to track everything, isolate experiments to branches, and separate your images to different tracks. Last, collaboration and benefitting from upstream is easier - you can take an image you agreed the most with, and just maintain your own changes, as well as in tracing where any issues occurred since you can rollback to previous 90 days on your builds and your upstreams.
That last one is why the base images deprecation alarms me so much. I don’t want to maintain my own - I agree with most of ublue’s decision and would rather alter a few things than setup the whole image myself. I may not be immediately affected, but the direction and speed of things make me worried that a major motivation for my use of uBlue is threatened.
It was fine, at least back then when I used GNOME. I just needed to use Kvantum and then things look and work fine.
My issue with the GNOME stack (as much as I don’t like talking about it too much, but I want it to be clear) is mainly with the functionality of it. CSD is a major annoyance - I have 768p laptop and ROG Ally, screen real estate is a premium, which is why I have my Unity-like setup. I want Menubar and Global Menu because it works - it can be on my top panel, but barring that, navigating with both keyboard and mouse is simple. On everything that isn’t GTK4, we have SSD, meaning expected behaviour like being able to Shade window, hide the titlebar automatically or manually (and, remember, Menu functions can be shifted to Global Menu), setting order of window buttons, and controlling the window from titlebar (moving Virtual Desktop by scrolling and customized left/right/middle clicks) all just works. Also, issues like cursor size suddenly being bigger on GTK app for some reason, is still not fixed.
I have, at this point, reached the point where I would rather have electron apps than GTK. And it all goes back to CSD because of course it does. I wouldn’t have that much problems if GTK/libadwaita isn’t so prevalent that it is the de facto default toolkit for Linux apps, and GNOME controlling it while not caring about designs that don’t fit their vision of how the desktop should be.
I won’t deny that there are issues with Qt as well, I have far more hope of it being fixed than Gtk.