koa
Experimental modal text editor built for extensibility and flexibility
modal text-editor
Star 8
Fork 0
Watch

fix(buffer): byte-based cursor columns collapse multi-byte UTF-8 #4

Open amartin.beer
Open opened by amartin.beer

Summary #

cursor-position counts bytes in :pre, so columns inflate for multi-byte UTF-8 (box-drawing, accented characters): the rendered cursor drifts and cursor-left/cursor-right can split a character.

Current behavior #

  • buffer.janet cursor-position carries a FIXME:; docs/runtime.md lists it under "Known tradeoffs".
  • ui.rs already maps byte -> char for rendering; the Janet side is byte-oriented.

Proposed approach #

Make columns char- (or grapheme-) aware across cursor-position, cursor-left/cursor-right, the move-lines column clamp, and future column memory.

Scope / notes #

First surfaced by the multibuffer separator glyphs (since removed from the buffer text), but user-typed UTF-8 still triggers it, and maybe emojis will shit the bed too.

Spork has an UTF8 module, but it's not pure, self-contained Janet code which we can yank like I did with the regex.janet module. JPM is not really embeddable, I will have to look for a good solution for this shitfest. I hate UTF-8.

Sign up or Sign in to add to the discussion.