Re: Unicode 3.1: IDS and ZW(N)J

From: John Cowan (
Date: Wed Jan 24 2001 - 14:44:01 EST

John Jenkins wrote:

> [B]ut in
> any event the basic stance we've taken in the past has consistently been
> that there is no expectation that IDSs will *ever* be rendered as single
> glyphs.

That is not what the Unicode Standard 3.0 says. It says that an IDS
may be rendered either as a sequence of glyphs or as a single glyph.

Allowing ZWNJ will prevent such rendering; allowing ZWJ will encourage
such rendering in fonts which can do it.

This actually doesn't require any magic processing during rendering,
except to add IDS sequences with ZWJ in them to font ligaturing tables,
as Unicode 3.1 now requires in joining scripts (Arabic, etc.)
It is more a matter of not forcing conformant IDS processors to treat
a sequence with a joiner in it as two separate sequences.

> Even allowing that smart font technology might make it possible
> to render them with single glyphs, I think they should be processed in
> the same fashion as (say) combining character sequences. We don't allow
> for a joiner between an e and an acute accent. If the system can render
> that CSS as a single glyph, then it can and that's as much as need be done.

Granted. But the word "ligature" is not used there, whereas it is
explicitly so used on page 271.

There is / one art             || John Cowan <>
no more / no less              ||
to do / all things             ||
with art- / lessness           \\ -- Piet Hein

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:18 EDT