-
Notifications
You must be signed in to change notification settings - Fork 22
Description
Currently we have only extremely focused benchmarks. dense_cells
tests a scenario where every single cell takes a huge escape sequence, while light_cells
tests for scenarios where pretty much no escape sequences are present.
When it comes to improving performance with something like alacritty/alacritty#8347, it's not always possible to improve every individual component of the parser. Since the "medium workload" in between dense_cells
and light_cells
is likely the best representation of what users would encounter in the real world, it should make sense to add this extra benchmark as a hint at real-world performance.
Generally speaking this is likely best simulated by an escape sequence heavy application, that doesn't just focus on generating graphical effects (like mpv -vo tct
). The best choice that comes to my mind would be vim with a workflow that includes both local edits, complete buffer jumps, and popups like fuzzy finders.