From: Gregg Reynolds (unicode@arabink.com)
Date: Tue Aug 02 2005 - 00:08:53 CDT
Kenneth Whistler wrote:
>
> And you haven't thought through the consequences of having duplicated
> digits with different directionality. You might think an end
Ok, here's an example. Assume that
123 are LTR, MSD
xyz are RTL, LSD (1, 2, 3, respectively)
and that lower case is LTR Latin, and Upper case is RTL Arabic.
Then "I saw one hundred twenty three elephants" looks like:
A) LTR context
display: i saw 123 elephants # LTR within LTR typesetting
coded text: [i saw 123 elephants] # and MSD polarity
B) LTR context
display: i saw xyz elephants # RTL within LTR typesetting
coded: [i saw zyx elephants] # but LSD polarity - bidi needed
C) RTL context
display: STNAHPELE xyz WAS I # RTL within RTL typesetting
coded: [I SAW zyx ELEPHANTS] # and LSD polarity
D) RTL context
display STNAHPELE 123 WAS I # LTR within RTL typsetting
coded [I SAW 123 ELEPHANTS] # and MSD polarity - bidi needed
(I tried to construct this using Arabic, but naturally found it
impossible to mix letters and digits.)
Obviously cases A and C are the most desirable; no string reversing
logic required. In each case the typographic semantics result it the
correct display.
As for existing software handling the new characters: every Unicode
character has typesetting semantics - directionality, combining, etc.
So when a new character comes along, applications that don't understand
it should not typeset it. Indeed it's a bug if they do, since you can't
typeset a glpyh if you don't know its typesetting semantics.
So to me that doesn't look so bad. What am I missing?
-g
This archive was generated by hypermail 2.1.5 : Tue Aug 02 2005 - 00:11:34 CDT