- From: Richard Ishida <ishida@w3.org>
- Date: Wed, 10 Jul 2013 20:42:41 +0100
- To: John Hudson <tiro@tiro.com>
- CC: public-i18n-indic@w3.org
Hi John, On 10/07/2013 19:32, John Hudson wrote: > On 10/07/13 6:55 AM, Richard Ishida wrote: > >> I put together some tests related to first-letter support. You can find >> them, and results for several browsers at >> http://www.w3.org/International/tests/html5/css3-selectors-first-letter/results-first-letter >> > > Thanks, Richard. These are very helpful tests. > > You write on the test site > > None of the browsers support selection of conjoined > consonants. These are basic units in indic scripts, and > all component characters need to be highlighted together, > but this goes beyond the definition of a grapheme cluster > in the Unicode Standard. > > It does? Perhaps because I'm coming at this from the font display and > Indic glyph processing side, I'm used to thinking in terms of the > graphic syllable (which does not always correspond to the linguistic > syllable) as the core unit of Indian writing, and presumed that this > corresponded to the grapheme cluster. Indeed, we regularly use the term > cluster as a synonym for graphic syllable. 'Grapheme clusters' in the Unicode definition of the phrase, include only the base character and the virama in a conjunct sequence, unfortunately. It's something I asked about a long time ago, and Mark Davis gave me a reason, of which I have now forgotten the details. I think it would be worth reinvestigating at some point though. > > Isolating the sequence of characters that make up a graphic syllable is > a standard first step in Indic layout; in the OpenType model, this is > performed by the external layout engine. It seems to me that the same > models -- indeed, the same code? -- could be applied fairly easily to > selecting 'first-letter' groups in browser support for CSS3 Selectors. Yes, that does indeed sound worth considering, and perhaps suggesting to the authors of the selectors spec. RI > > JH > > -- Richard Ishida, W3C http://rishida.net/
Received on Wednesday, 10 July 2013 19:43:10 UTC