Re: Proposal: Ligatures w/ ZWJ in OpenType

From: Doug Ewell (
Date: Tue Jul 16 2002 - 11:21:20 EDT

[OpenType address removed.]

John Hudson <tiro at tiro dot com> wrote:

> Perhaps the idea is that, by keeping things vague, the UTC permits
> freedom of implementation, but so far all I am seeing in response is
> confusion: confusion about what ZWJ signifies in text, and how it
> should be implemented in line layout. If Doug is worried that ZWJ will
> be 'deprecated on arrival', he might also worry that ZWJ will be so
> variously interpreted as to become useless as a reliable means of
> achieving any consistent result.

I don't see any confusion in the interpretation of either ZWJ or ZWNJ.

ZWJ requests the "most connected" rendering possible of the two (or
three...) adjoining characters, including ligatures if possible; ZWNJ
requests the "least connected" rendering possible.

If a ZWJ appears between two characters, but the current font does not
include such a ligature, the two separate glyphs *are* the most
connected rendering possible.

In the case of e x p a n d e d text (where the expansion is performed
by font settings or line layout, instead of brute-force spacing as I
just used), it's true that an ill-placed ligature may look really bad.
This has little to do with ZWJ, though. The rendering system must
decide when to ligate and when not to, regardless of whether ligation is
requested by ZWJ, by markup, or by an application's global setting.

-Doug Ewell
 Fullerton, California

This archive was generated by hypermail 2.1.2 : Tue Jul 16 2002 - 09:34:03 EDT