You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Following up from #3506 and #3438 (comment), when a label is vertically oriented with a vertical writing mode, any one- or two-digit numbers (and one- or two-letter words) would ideally be left upright, their digits placed side-by-side. The same should happen for one- or two-letter runs of Latin text. This effect is called tate-chu-yoko in Japanese:
Currently, we replace a punctuation character with its fullwidth analogue if both adjacent characters are compatible with vertical writing. This heuristic is conservative enough that we could implement single-character tate-chu-yoko simply by mapping all the ASCII numbers and letters to their fullwidth analogues in lookup (or leftAndRightBinding once #3587 is implemented), or by adding 0xfee0 to the character code of any ASCII number or letter.
Double- or triple-character tate-chu-yoko would require the special logic to avoid rotating ASCII characters, position them within the correct half-character box, and use specialized characters like ㏾. While impressive, I think anything beyond single-character tate-chu-yoko is unnecessary for the time being.
Following up from #3506 and #3438 (comment), when a label is vertically oriented with a vertical writing mode, any one- or two-digit numbers (and one- or two-letter words) would ideally be left upright, their digits placed side-by-side. The same should happen for one- or two-letter runs of Latin text. This effect is called tate-chu-yoko in Japanese:
Currently, we replace a punctuation character with its fullwidth analogue if both adjacent characters are compatible with vertical writing. This heuristic is conservative enough that we could implement single-character tate-chu-yoko simply by mapping all the ASCII numbers and letters to their fullwidth analogues in
lookup
(orleftAndRightBinding
once #3587 is implemented), or by adding0xfee0
to the character code of any ASCII number or letter./cc @lucaswoj @nickidlugash @friedbunny
The text was updated successfully, but these errors were encountered: