it is based on Andrea Corallo’s
feature/native-comp branch, which contains these small modifications so far:
- Process IO modification: together with IO-related settings described below, on my computer
shellcan cat large files 60x faster than original emacs. Working with large log files is much easier this way. Emacs is still slower than fast terminals, but its speed is much more acceptable.
- Rendering optimization: face lookup by face name is done by hash table instead of linear search. This makes display rendering much faster in certain cases (here’s such a case: when I had the same buffer open in several windows, and in evil mode, I switched to visual mode to select some text, emacs slowed down to ~2 fps. With this modification, slowdown is almost unnoticable).
- Disable composition. As I don’t use ligatures nor fonts which need composition, I disabled this feature. If you’d like to still use composition, reverse the related commit (it is just 1 line).