From: Philippe Verdy (verdy_p@wanadoo.fr)
Date: Fri May 14 2004 - 18:37:29 CDT
From: "Mark E. Shoulson" <mark@kli.org>
> I didn't say it couldn't possibly done. But it IS too complex a
> situation for raw Unicode to handle, in general. Considering how weird
> some results come out with the normal bidi algorithm as it is,
> boustrophedon not something that should be handled in plain text, and
> that's about all there is to it.
I don't see any place in boustrophedon scripts that forbids them to work with
BiDi. one just needs to make it work like BN (Bidi Neutral) and affect their
direction from the previous character or the occurence of a line break (which
inverts its current direction, but still keeps it in a BN context. All the
complex cases are already resolved with scripts having strong directionality.
Boustrophedon script is a weaker and allows more flexibility than in the case of
mixed (strong RTL + strong LTR). Small inclusions of Boustrophedon in either RTL
or LTR texts will be mostly transparent (unless there's a line break in the
middle of this boustrophedon fragment).
Most boustrophedon scripts can also be handled in a strong RTL or strong LTR
style, where the standard BiDi algorithm can be used. For Archaic Greek for
example, its inclusion within Hebrew texts would adopt the Hebrew
directionality, but in a Latin text, it would adopt the LTR directionality. For
this example, the strong directionality of the Greek script with which it is
unified is too strong. What should be noted is that directionality should be
made weaker than in the existing standard. Shamely there's no Bidi override
control that allows setting a weak (contextual) directionality. Boustrophedon is
a specialized subcase, but there are scripts that can be written indifferently
from LTR or from RTL. With a weak directionality, we could avoid direction
breaks when rendering lines (meaning that all glyphs representing characters
with weak directionality would NEVER need to be reordered).
This is a case where a BDW (Bidi-Weak Override) control could be helpful for
sections of text that should be rendered with weak directionality (for example
with inclusions of Archaic Greek terms within semitic texts).
This archive was generated by hypermail 2.1.5 : Fri May 14 2004 - 18:38:00 CDT