Hyprland 0.55 announced the switch to Lua for its config files

The practical defense of Lua was strongest for window managers, terminals, editors, and similar power-user software. People want machine-specific settings, event-driven behavior, layout logic, and tight synchronous reactions to state changes. Without an embedded language, that logic does not disappear. It leaks into shell snippets, sidecar daemons, IPC clients, YAML generators, or template systems. Several comments argued that Hyprland’s old config was already drifting in that direction, so moving to Lua is less a philosophical shift than an explicit acknowledgment of what users were already trying to do. The case against it was just as concrete. Once config becomes code, it gets harder to reason about, easier to overengineer, and more fragile across upgrades. Build systems were the warning example. Gradle and Nix came up as proof that programmable config can become a maintenance tax fast. People who prefer Sway, i3, Ghostty, or niri liked the separation better: keep the core config readable and stable, then push real logic into external programs, include files, or companion daemons with smaller integration surfaces. On Hyprland specifically, the mood was mixed but not shocked. Some people said migration is easy, converters already exist, and simple personal configs rewrote cleanly. Others said Hyprland has a history of config churn and fast-moving dependencies, so another config reset reinforces the sense that it is still a moving target. The strongest consensus was not "Lua good" or "Lua bad." It was that configuration systems almost always drift toward programmability, and the real design choice is where you want that complexity to live and how visible you want it to be.