text run segmentation problems

[christianparpart]

trafficstars

Example screenshot:

Reproduction snippet:

echo -e "\e[31mabc\e[32m...\e[m\nabc...\n"

Text run segmentation properties:

Notes

Text run egmenation should happen based

• SGR attributes
• word delimitation (whitespace)
• script property
• emoji presentation property

A single text sub run contains has each of these properties fixed for a sequence of codepoints.

So in theory for the above example "abc..." should be one single sub-run (because "Common" script will be merged into "Latin" script)

Also, I could not find another case that has similar issues to this. It seems to only affect this case (AFAICT) with some text followed by the "..."-ligature using "JetBrainsMono" (or Fira Code, others?).

Maybe first validate if that holds true for the above reproducible case by looking at the debug output for debug tags font.*