Re: [css-ruby] spanning of ruby annotations across excess bases

On Wed, Nov 12, 2014 at 9:20 AM, kawabata taichi <kawabata.taichi@gmail.com>
wrote:

> Dear David,
>
> On Sun, Nov 9, 2014 at 3:42 AM, L. David Baron <dbaron@dbaron.org> wrote:
>
>> http://dev.w3.org/csswg/css-ruby/#base-annotation-pairing says:
>>   # If there are not enough ruby annotations in a ruby annotation
>>   # container, the last one is paired with (spans across) any excess
>>   # ruby bases. (If there are not any in the ruby annotation
>>   # container, an anonymous empty one is assumed to exist.)
>>
>> Is there actually a use case for this behavior, or is it really just
>> defining error handling?
>>
>
> As far as I know, this is an intended behaviour based on existing use case.
> JLREQ usage (Fig. 3.61) usage is certainly rare, but it does have non-
> ignorable
> presence in various occasions.
>

Even if we drop the spanning support here, it can still be achieved with
nested ruby. Given that it is rare, and is usually used only on a single
word which is not surrounded by other ruby boxes, I think using nested
markup instead is an acceptable solution.

Also, it copes with HTML5 Ruby behaviour.
> cf.
> http://www.w3.org/TR/html5/text-level-semantics.html#annotation-pairing
>
> - When ruby-base elements excesses, all of them matched into last of
> existing ruby base.
> - When ruby-text elements excesses, it matches to virtual empty ruby texts,
>

Yes, we have that in W3C HTML5, but we don't have it in WHATWG HTML5. I
think the behavior defined in W3C HTML5 is just a simplified version of
that in Ruby Anotation spec for XHTML http://www.w3.org/TR/ruby/ . Given
the use cases proposed in XHTML spec, I think nested ruby is even better in
semantics for many of them.

- Xidorn