Dirty "regions" instead of dirty lines?

[davidbrochart]

The dirty screen attribute is a set of line numbers which should be re-drawn. I'm using pyte in a Textual application where one can decide to update a region, where a Region is a rectangle.
I can of course say that for each dirty line corresponds a dirty region that has the width of the screen and a height of 1, but it would be more optimal if I knew what (segments of) characters are dirty inside each dirty line. I'm wondering if pyte could have something similar, i.e. dirty regions instead of just dirty lines?

[superbobry]

Yeah, you can fork DiffScreen and implement your own version tracking dirty regions. IIRC we went with lines because it was simpler and sufficient for our use-cases.