W3C home > Mailing lists > Public > www-style@w3.org > May 2012

RE: [css3-text] feedback on 'word-break: keep-all;'

From: Koji Ishii <kojiishi@gluesoft.co.jp>
Date: Mon, 7 May 2012 09:40:58 -0400
To: "Kang-Hao (Kenny) Lu" <kennyluck@csail.mit.edu>
CC: Ambrose LI <ambrose.li@gmail.com>, fantasai <fantasai.lists@inkedblade.net>, WWW Style <www-style@w3.org>
Message-ID: <A592E245B36A8949BDB0A302B375FB4E0D3C3C80B5@MAILR001.mail.lan>
> 1. It preserves breaking after hyphens.

Do you mean U+002D? It's HY as per LineBreak.txt[1], and therefore not affected by this value.

> 2. It preserves breaking after fullwidth punctuation (and so it doesn't just break after word
> separators).

Depends on which fullwidth punctuation you're talking about. U+3002 IDEOGRAPHIC FULL STOP for instance is CL, so it's not affected either.

> 3. It no longer breaks between an ideograph (including Korean
> characters) and a latin character (AL/ID, ID/AL). This is somewhat surprising to me, but
> IE9 already implements this. Perhaps it would be worth checking with the Korean Interest
> Group again to see if that's desired.

IE has been implementing this[2] since IE5 (or 5.5, I don't remember,) and this property is already used in web pages for a decade or more. The goal is to standardize it in CSS Text Level 3 for other browsers to implement.

There could be some punctuation where the current IE implementation does more, I haven't run through, but recent discussion at public-html-ig-ko indicates that this value isn't used that often, so I think following UAX#14 is a good approach.

> I have one remaining question which is also relevant to Ambrose's question. If a browser
> is able to determine to some extent the word boundaries of a run of Chinese or Japanese
> (it seems that WebKit browsers can do this, with certain failures of course, for double click
> word selection), can 'word-break: keep-all' make it break at these points?
> 
> I think this is probably too advanced and perhaps difficult to spec at this level, although I
> am interested in what people think.

Could be interesting, but its quality is pretty low as you pointed out, and it's hardly interoperable, so I don't think authors would want to rely on such feature. It's interesting to know WebKit implemented something though.

[1] http://www.unicode.org/Public/UNIDATA/LineBreak.txt

[2] http://msdn.microsoft.com/en-us/library/ms535149(v=vs.85).aspx

Regards,
Koji
Received on Monday, 7 May 2012 13:42:14 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:53 GMT