- From: John Daggett <jdaggett@mozilla.com>
- Date: Mon, 17 Jan 2011 17:50:52 -0800 (PST)
- To: MasaFuji <masa@fuji.email.ne.jp>
- Cc: www-style list <www-style@w3.org>, Thomas Phinney <tphinney@cal.berkeley.edu>
Masahiro Fujishima wrote: > I have an understanding of typographical fear of risks. But for CJK > fonts, it is very useful and economical tactic to use > expanding/condensing method. As you know, CJK font has more than 3,000 > or 5,000 characters in it. It is impossible to prepare ideal series of > font-width in a font family. There was a discussion along the same lines back in August 2010: http://lists.w3.org/Archives/Public/www-style/2010Aug/0267.html The one key point to me is that "effects" like this should be clearly separated out from the font selection process. So 'font-stretch' should control *selection* and a separate property (e.g. 'font-xscale') should control artificially synthesized condensed/expanded. Bolding works automagically but that's an unfortunate feature left over for compatibility reasons. Note that you can already use CSS transforms to achieve the same effect: http://people.mozilla.org/~jdaggett/tests/xtransform.html Using transforms for this is somewhat clumsy but possible nevertheless. Note that for body text sizes, using fake condensed/expanded glyphs isn't going to work in an ideal way in most implementations, since to really be correct the font rasterizer needs to be involved, it needs to do the stretched/shrunk rasterization; just applying a simple image transform on a normally rasterized glyph will produce incorrect subpixel antialiasing. Not such a big deal for display sizes but definitely not ideal for text sizes, especially in the case of high information density glyphs characteristic of CJK fonts. So my thought would be to consider a separate property in the future but not for inclusion in CSS3 Fonts. John Daggett Mozilla Japan
Received on Tuesday, 18 January 2011 01:51:26 UTC