- From: fantasai <fantasai.lists@inkedblade.net>
- Date: Sat, 27 Jun 2015 07:41:43 -0400
- To: Xidorn Quan <quanxunzhen@gmail.com>, www-style list <www-style@w3.org>
- CC: Masayuki Nakano <masayuki@d-toybox.com>, Koji Ishii <kojiishi@gmail.com>, Shinyu Murakami <murakami@vivliostyle.com>
On 06/26/2015 08:25 PM, Xidorn Quan wrote: > Hi, > > Recently, Masayuki mentioned that, for Japanese, if a piece of > text in vertical writing mode has both ruby and underline, the > underline should be outside ruby annotation, like: > > 分わ| > か | > る | > > However, implementors who are not aware of this will tend to implement that as: > > 分|わ > か| > る| > > (This is actually the current behavior of Firefox Nightly) > > I know the current spec says the exact position of underline is > UA-defined, but I guess it at least worths a note. Actually, I > don't think this is a behavior could be covered by "UA-defined". > > As an implementor, I feel it looks hard to implement, but probably > not something impossible. > > Also it seems to me this is not mentioned in JLReq. I guess we > should also cc people there and ask their opinion about this. Good point. I agree with asking the JLTF to put a clarification into JLREQ--for this situation in both horizontal and vertical text--and we should follow up by figuring out a good model for handling this in CSS. I'm not 100% sure what that should be, as there might also be cases where the underline should go between the ruby and the base. At the moment, I lean towards saying that underline specified directly on a ruby container should be under the lowest level of text, overline should be over the highest level of text, and line-through should be applied to the base level. Decorations specified on a descendant of the ruby container should of course only apply within that descendant's level. But for ancestors of the ruby container, I am less sure what the correct answer should be. ~fantasai
Received on Saturday, 27 June 2015 11:45:06 UTC