RFC 2 · How should mixed CJK and Latin content behave in the same column?

[jihoons0]

Labels: RFC typography

A single reference like 「창 1:1」 contains hangul, an ASCII colon, Latin digits, and CJK brackets — each with different orientation rules. Unicode (UAX #50) defines orientation per character class, but browser and OS implementations vary, and tate-chu-yoko (縦中横) grouping of short digit runs is inconsistent.

Where it shows up here: verse, text-field, and Vertically News under live-headline load.

Open questions:

• Which digit/acronym runs should auto-compose as tate-chu-yoko, and where does that break down?
• How do we keep punctuation upright vs. rotated consistently across engines?
• Is there a token-level way to declare per-class orientation so component authors don't re-solve it each time?

Write-up: https://vertically.works/challenges#mixed-language
Propose your answer as a PR. Strong answers get merged into the spec.