@janl
Maybe like scrolling window manager ( #niri) vs. classic tiling ( #sway) vs. classic stacking (#GNOME/#KDE).
@75watt @eschaton

@janl @eschaton

Think desktop vs mobile, or a very light tiling desktop vs a full featured one like KDE Plasma, each can implement the parts of the protocol they need.

The idea was to avoid creating again a huge unmaintainable bloated mess like what the X server had become.

@janl modularitiy is a core idea here.

This architecture makes it possible for anyone to create a piece of the system and integrate it seamlessly into the existing infrastructure.

How requirements may differ in the future is (at least near) impossible to predict, but _that_ requirements change is almost certain.

This way, the system can be adapted and users can pick the module they like best.

@janl @75watt My guess is that it’s a theoretical benefit, just like X11’s insistence on “mechanism, not policy” was a theoretical benefit that was actually actively harmful to the X11 software ecosystem by ensuring that there was no such thing as a standard UI toolkit unless you count either Athena widgets (part of the distribution) or Motif (shipped by every major vendor prior to 386BSD and Linux).