Re: [css-ruby] Default font-size of ruby annotation with ruby-position: inter-character

On Tue, Jan 27, 2015 at 1:13 PM, John Daggett <> wrote:

> Xidorn Quan wrote:
> > OK, then, let's add a keyword for 'font-size', say 'ruby-text', and
> > make it be computed according to ruby-position. It is computed to 30%
> > when ruby-position is inter-character, and 50% otherwise.
> Adding keyword values like this to 'font-size' affects the parsing of
> the 'font' shorthand so it's a fairly large impact change.

Hmm, but we have had several keyword values for font-size, and the 'font'
shorthand has a relatively strict syntax requirement, which actually allows
us to introduce any type of single value for font-size without causing

> The default
> stylesheet font-size settings for ruby elements should cover the
> majority of use cases. How frequently do you expect 'inter-character'
> will be used in practice? The ruby spec describes it as something
> that's a special case.

Less frequently than ruby in Japanese, I guess. Maybe Bobby can answer this

> I'm also suspicious that as to how "fixed" the
> percentages you've listed above actually are, especially across common
> use cases in Japan, Taiwan and China. If the common usage is slightly
> different in Taiwan vs. Japan for example then I think the fixed
> nature of 'ruby-text' won't be a win.
> I should also point out that house rules for typesetting like this
> often vary. The CLReq or the Taiwanese manual you linked to may state
> a single value but I suspect that's not written in stone and
> publishers would vary it based on context (e.g. larger ruby sizing for
> text aimed at younger children).

I don't think Taiwan publishers would often enlarge the size because that
would cause the annotation exceed the boundary of one character, which I
think rarely happens.

But it makes me wonder whether ruby-text should be computed according to
ruby-position. It's probably true that inter-character will only be used in
Taiwan, however it is not true that Taiwan people only use inter-character.
They may use other values as well, and in those cases, the font-size should
also be 30%, because it is the usual font-size for bopomofo.

Maybe the best way is not introducing a new keyword, but specifying a
locale-dependent default stylesheet.

> The other issue is that on the web, sizing rules like this are
> distorted because of the discrete nature of smaller sizes for text
> displayed on screens. Where in printing a consistent ratio of text
> size to ruby size might be used, for screens it's common to bump up
> the relative ratio at smaller sizes. This is how superscripts and
> subscripts are displayed in HTML for example.

It might be an existing problem for the current spec. I don't think adding
the new keyword would make anything worse on this.

- Xidorn

Received on Tuesday, 27 January 2015 03:16:15 UTC