- From: John Hudson <tiro@tiro.com>
- Date: Thu, 13 Mar 2014 16:22:17 -0700
- To: "Cibu Johny (സിബു)" <cibu@google.com>, Richard Ishida <ishida@w3.org>
- CC: John Cowan <cowan@mercury.ccil.org>, indic <public-i18n-indic@w3.org>
On 13/03/14 2:35 PM, Cibu Johny (സിബു) wrote: > It's boundary inside a grapheme cluster is decided by an explicit virama > in its shaping. This is where the character-based determination of a graphemic entity ceases to be sufficient, though, because different fonts may display the same conjunct with or without explicit (visible) virama; indeed, the same font may display the same conjunct in different ways depending on language system tagging (e.g. a conjunct may shape as a stacked ligature for Sanskrit, but with an explicit virama for Hindi). Because of modern rules for ordering ikar and reph relative to explicit virama, the OpenType Indic specs include different conjunct ligature shaping features for those involving true half forms and those involving explicit virama, and reordering takes place between application of the two features. Something similar could be done in terms of isolating Cibu's 'explicit graphemes', but it means interacting with the layout engines and fonts, which strikes me as an incredibly complicated way of handling CSS first-letter styling. JH
Received on Thursday, 13 March 2014 23:23:33 UTC