The Unicode Consortium Discussion Forum

The following code points:

3031 VERTICAL KANA REPEAT MARK
3032 VERTICAL KANA REPEAT WITH VOICED SOUND MARK

are cl-19.3/U in the current draft.

Our recommendation is cl-15. cl-15 and cl-19 differ in ruby overhang, and these two code points should allow ruby overhang.

Also, although these code points were defined for vertical text flow, for systems that can switch text flows, we'd like to allow font vendors to design horizontal glyphs to put in to default glyph, so we think T is more preferable, but technically speaking that difference won't matter much, so I'm not insisting much on that.

If we're setting them as T, we'd also recommend to set the following code points to T as well:

3033 VERTICAL KANA REPEAT MARK UPPER HALF
3034 VERTICAL KANA REPEAT WITH VOICED SOUND MARK UPPER HALF
3035 VERTICAL KANA REPEAT MARK LOWER HALF

These code points can be treated in the same way as Mongolian, since glyphs in the code chart shows vertical glyphs.

In tradition, these code points were never used in horizontal flow and therefore assumed to be used in vertical flow only, so no one has good idea how to render these code points in horizontal flow. But everyone also would agree that using the glyphs in the code chart in horizontal flow is incorrect. In that sense, I see similarity with Mongolian.

JLREQ authors have come up with the most reasonable glyphs for these code points in horizontal flow. I couldn't find how I can attach image files here, but I can provide one if it helps.

Whether their idea is correct or not is controversial, but given everyone I talked to thinks using the glyph in horizontal flow is incorrect, from UTR#50 point of view, these should be marked as special, just like Mongolian cases.

I don't mind changing them to cl-15.

After that change, would it be the case that allowed ruby overhang could be entirely derived from the EAC property?



On that basis, I'd rather not move them to T, as this could be perceived as taking a stance, such as "blessing" their usage in horizontal text. I'd prefer to deal with the situation via the text (just like Mongolian).

Eric.

JLREQ classes are designed to be used in line breaking, spacing, and ruby overhang, so the simple answer is yes, but you need to pay attention to a few caveats:

• JLREQ is a requirement document and tries to show the concept in a simple way, and therefore it hides some minor details as you know
• There are some variations for ruby overhang rules. JLREQ classes support simple case, and cases like jukugo-ruby in JLREQ are not covered

That's a possible option, but we probably want to recommend font vendors to implement somewhat usable glyphs in horizontal flow, don't we? I hope you agree that "this text isn't used in horizontal flow" can no longer work today, and therefore I'd like to take these a bit different from Mongolian.
If you want text solution, that's okay, but different text might be needed than Mongolian.

Given the refocus of UTR#50, I have addressed this issue at the end of section 4: