handle gestures and prediction at the same time
input-method: handle text from the keyboard
input-method: use std i/o instead of a child
Add manager script
Add a license (MIT)
Support both zig 0.8 and 0.9
Add simple completion reference implementation
Add wayland protocol specifications
input-method: send selected text to the compositor
selector: use frame callbacks
selector: Cleanup memory we allocate directly
selector: communicate with stdout / stdin
selector: wire state to dynamically change words
selector: use default values in structs
selector: start building state for storing options
selector: use pango / cairo instead of setting pixels
selector: use layer-shell instead of xdg
build: compile both executables
input-method: Pipe current word to a predictor