Re: Order of bidi script numbers in ranges

From: Roozbeh Pournader (
Date: Fri Jan 19 2001 - 13:07:10 EST

On Fri, 19 Jan 2001, James E. Agenbroad wrote:

> In what order are ranges of numbers such as 15-23 expressed in a bidi
> context? 1. What is wanted visually, if there is one consistent
> expectation

It's really locale dependent. It's used differently even with two
publishers with same audience. The style guide at my hand (published by
Iran University Press), tells:

        The dates of people's birth and death are written from right to

which is the nearest thing to your question. It then gives an example for
Gauss that's visually "1855-1777". But if you ask about my own experience
with Persian texts, I've seen more "1777-1855"s.

> 2. Then what order should the codes be stored in Unicode for
> the bidi algorithm to provide the desired visual order? My guess is that
> the visually the '15' would be wanted to the right of the dash and '23'
> and that this would be the desired order in a Unicode string too, but it's
> only a guess. Thus the Unicode test string would contain:
> 1. The first string of Arabic or Hebrew characters
> 2. The code for '1'
> 3. The code for '5'
> 4. The code for dash
> 5. The code for '2'
> 6. The code for '3'
> 7. The code for remaining Arabic or Hebrew text.
> Apoligies if this is obvious to everyone or explicit in 3.0.

The logical order of characters is right, as this is the
machine-processable order, and also the order the user types the text,
but you may need to insert some bidi marks here and there to to handle
different locales.

> May I assume
> decimal numbers such as 3.1416 and time such as 10:30 are expressed in the
> order used in the West (though the punctuation may differ)?

Yes (at least for Persian and Arabic locales).


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