Am Mittwoch, 4. August 2010 um 08:52 schrieb William_J_G Overington: WO> Please know that, whilst I comment on various matters, I am WO> enthusiastic for the general thrust of your suggestion regarding WO> access to alternate glyphs for Latin characters using Variation WO> Selectors. This could produce a renaissance for typography.
Admittedly, I explicitly do not want to introduce glyph encoding into Unicode through the back door. In the next version of my proposal, you will find some words about what variation sequences are *not* intended for. WO> >> But while the general mechanisms for doing so are standardized WO> >> (i.e. OpenType features), the concrete selection of a specific glyph is not. WO> WO> It is important that the Unicode specification does not regard WO> any particular font technology as being the standard font technology. This is correct. I mention OpenType only as an example. WO> Why is it not possible specifically to request a one-storey form of lowercase letter a? I did not this, as I do not know a cultural context where the two-storey form is to be suppressed to prevent an "a" to be mistaken for any letter too similar to a two-storey a. WO> What happens in relation to a character such as g circumflex? WO> Would one be able to access a glyph alternate for g circumflex? The variant selector can be followed by any diacritic which then is applied to the base character. WO> Could there be variants for lowercase e, ... I have found none, which of course is no proof of non-existence, WO> for a horizontal line glyph design, and for an angled line, Not according to the principles outlined in my proposal, WO> Venetian-style font, glyph design please? No. WO> Would it be possible to define U+FE15 VARIATION SELECTOR-16 to WO> indicate an end of word alternate glyph for each lowercase Latin WO> character? No. Even if you find a cultural context where such things are required, such things are positional variants which are to be handled by the proven mechanisms developed for scripts like Arabic. - Karl Pentzlin
