- From: John Hudson <tiro@tiro.com>
- Date: Thu, 18 Nov 2010 09:12:29 -0800
- To: www-style list <www-style@w3.org>
- CC: Joel Kalvesmaki <KalvesmakiJ@doaks.org>, Ross Mills <ross@tiro.ca>
Re. http://dev.w3.org/csswg/css3-fonts/ In thinking through implementation of Byzantine sigillographic transcription using CSS3 font module properties -- which is turning out to be a great test case --, I came up with the following situation, and want to check whether this is addressable with the proposed CSS syntax for character variants and ligatures. The situation is that a sequence of characters, e.g. the uppercase Greek letters ΚΑΙ (Kappa Alpha Iota), may be represented by a ligature glyph /Kappa_Alpha_Iota/, which is implemented in the font using the Discretionary Ligatures <dlig> feature and in CSS using the font-variant-ligatures property with the additional-ligatures value. But the ligature representing ΚΑΙ may take two different forms, and the font provides a variant glyph /Kappa_Alpha_Iota.2/, which is mapped as an enumerated variant in one of the font's Character Variant features, say <cv60>, implemented in CSS using the character-variant property. Note that this is a variant of the ligature form, not a variant of the underlying letters, i.e. the ligature formation precedes the variant selection. The variation in ligature form is not dependent on the form of the underlying characters, i.e. it is a free variant of the ligature representation not influenced by e.g. the particular shape of the Kappa. Is there the potential for ambiguity in property interaction, i.e. might the character-variant property be interpreted as applying to the underlying letters rather than to the ligature? Is this something that is influenced by the ordering of CSS properties, or is it something that will be determined solely by the ordering of the GSUB lookups in the font? [In the case of this font, all ligature substitutions will precede any character variant substitutions. I can imagine other fonts in which the variant form of the underlying letters dictates the ligature form, so in that case the character variant lookups would precede the ligature lookups. I can also imagine fonts that use a combination of both approaches, with staggered lookups.] JH
Received on Thursday, 18 November 2010 17:13:05 UTC