W3C home > Mailing lists > Public > public-i18n-cjk@w3.org > January to March 2012

Re: Memo from ruby disucssion with Roland

From: fantasai <fantasai.lists@inkedblade.net>
Date: Wed, 22 Feb 2012 12:21:03 +0100
Message-ID: <4F44CF9F.4070307@inkedblade.net>
To: public-i18n-cjk@w3.org
On 02/22/2012 05:37 AM, Roland Steiner wrote:
> Sorry for chiming in late to this thread!
> Koji summed up our discussion pretty well, I think. There's just one thing I wanted to add to the reasoning of the 'for'
> approach for clarification:
> With 'rspan' or <rtc>, there is no semantic link from the base to the text(s). For example:
>     <ruby><rb>東<rb>京<rtc>とうきょう</ruby>
> and similar with 'rspan' - in this case, even if you add, remove or edit the bases dynamically, the ruby text will always be '
> とうきょう'. This is what I wanted to address with the 'for' attribute: the text would only be rendered if the bases it refers
> are present within the same ruby, and rendered in the order it refers them. E.g.
>     <ruby><rb id=T>東<rb id=K>京<rt for="T K">とうきょう</ruby>
> In this case, 'とうきょう' should only be rendered if there are the bases 'T' and 'K', consecutively, in that order. If, for
> example, the base for '東' is removed, the text would no longer be applicable, and thus not rendered. Neither would it be
> rendered if 'T' or 'K' are not the id of a <rb> within the same ruby.
> Now, I fully admit, that this is very verbose may not be all that necessary. It also doesn't help if the innerText of the
> bases is changed. Implementation is also tricky. OTOH, the advantage would be that a <rt> does not have to neighbor a base.

I agree it's very verbose, and I don't think it's really that necessary.
The assertion that there's no semantic link from the base text is
not true -- the association is implied by the structure of the markup.
We do this for tables, for <dl> markup, etc. As long as it's unambigously
defined what associates with what, there is no problem and it's perfectly

(I don't think your markup is correct, either -- Tokyo's annotation
is not group ruby; there should be two <rt>s, one for each base.)

You've seen http://fantasai.inkedblade.net/weblog/2011/ruby/

> The whole idea came from a discussion to alternatives to markup. That is, for some use cases it would be nice to be able to
> generate ruby texts by CSS rather than having to repeat markup. Along the lines of:
> rb.東::ruby-text {
> content: "とう";
> }

This out-of-scope for CSS: CSS is for style, not for content. It
should be possible to replace every phrase in the text with a
paraphrase and the style sheet should be equally appropriate to
the page. And it should be possible to suppress the style sheet
and not lose any content.

The phonemes proposal, for example, was dropped from CSS Speech
because it tied the style sheet to specific content. The right
solution there was to link the HTML file to a pronunciation

So, if out-of-band ruby annotations are needed, I suggest creating
a format optimized for that, rather than repurposing CSS. If you
like Selectors, though, you can re-use Selectors. :)

Received on Wednesday, 22 February 2012 11:21:35 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 22 February 2012 11:21:36 GMT