I think configuring software via editing the code is not a bad approach, but the suckless approach less flexible/composable than what XMonad does. In suckless

I think configuring software via editing the code is not a bad approach, but the suckless approach less flexible/composable than what XMonad does. In suckless software, you edit config.h, but then if the change isn't exposed in config.h, you need to modify the source code. In the case of XMonad, you can think of XMonad instead as a "library for writing window managers". You customize it by writing your own main function, and calling/combining the appropriate library functions [0]. This way, there's text-based patches to maintain; the library code is unmodified, all your changes are in your own file. [0] - There is a prebuilt simple main function that lets you customize the basics, similar to the suckless config.h