05:41 <dnkl> cbb: sterni: we could probably (ab)use the mechanism for grapheme clusters to implement the width part of the kitty protocol. That is, store the entire string, along with the number of cells it uses, in the first cell.

05:42 <dnkl> as long as it's only used for "problematic" characters (like multi-codepoint emojis), there won't be a performance impact. But if people start emitting it for *every* character... then it'll get really slow.

05:51 <dnkl> xaltsc: faster startup times, mostly. You'll also get a pre-filled glyph cache which might speed up rendering (at least initially) in some cases.

06:07 <rockorager> dnkl: fwiw my planned usage would be to only use it for multi-code point, ambiguous width, or PUA glyphs. I still wish kovid would accept my proposal to use an sgr for width so you can use this safely without detecting support

06:07 <rockorager> I think of those it’ll be the PUAs that see most usage with nerd font icons