3 ms·
You've said the trigger words for my pet rant/ theory, sorry ;) I dont know that the problem is balkanization on the front end: the window look and feel, etc.
by kjellsbells 4y ago
You've said the trigger words for my pet rant/ theory, sorry ;)
I dont know that the problem is balkanization on the front end: the window look and feel, etc.
There used to be one problem. Look and feel was horrible. Kind of the uncanny valley, where people coming from Windows or Solaris had to deal with a UI that was almost right but also wrong in dozens of tiny ways. And those were the good old days where all you had to do to switch was replace the path to your window manager in your X startup file.
Now we have two problems: the l&f is still weak across the board, with no stable UI/human interface standards nor a framework for designing themes that cover every part of the UI (ie users can skin their desktops and window chrome, but its always clunky, standards are low and even the best themes dont extend everywhere in the UI...shades of Windows' infamous Control Panel/Settings inconsistency here)
And secondly, changing desktop environments is incredibly ugly for users to do. Ripping out gnome for kde is fastest by simply reinstalling the OS. The hooks are so deep, the graphical decision you make when you download your install image is effectively immutable. Isnt it a little bizarre that fedora and ubuntu have to provide separate images simply because the desktop environment is different?
If I had a magic investment wand here, I would spend money on pro designers to spiff up the look and feel of a couple of desktop environments, like popOS does, and make them selectable at install time not download time. Id make switching trivially easy. And Id throw a boat load of money at defining a stable DE framework that resolved the kde gnome schism once and for all.
- mappu 4y ago> And Id throw a boat load of money at defining a stable DE framework that resolved the kde gnome schism once and for all. Even with a boatload of money, how would you go about this? Write a shim lowest-common-demoninator API that can target both (like wxwidgets)? Theme them to look like each other (like Qtcurve)? Extract more and more individual concepts that can be collaborated on and shared (like fd.o / xdg-shell)? I think the key problem is the proliferation of incompatible object extensions to C (C++ vs GObject and also Objective C). Porting GObject to C++ (like Gtkmm) is the most developed direction, the alternative of porting Qt to C/GObject (make all GObjects implement QObject or vice versa?) has not even been tried AFAICT.