Re: Arabic and character-cell terminals

From: Joan Aliprand (BR.JMA@rlg.org)
Date: Fri Jun 04 1999 - 14:16:38 EDT


>Frank da Cruz <fdc@watsun.cc.columbia.edu> asked:
>
>This question is related to Unicode, even though it might not seem to be.
>
>Are there now, or have there ever been, character-cell terminals with
>Arabic keyboards and that display Arabic text on the screen?

Yes.

>And if so,
>how are characters placed on the screen? --
>
> 1. All the same width (monospace).
>
> 2. Some normal (ASCII) width, others doublewidth (duospace).
>
> 3. Variable spaced.

When RLG added Arabic to our terminal emulation software, we knew
monospacing would not work. The comparative differences in the
width of Arabic letters are much greater than in Latin script.

Our terminal emulation software is a special-purpose application for
RLIN, the Research Libraries Information Network, that runs under
Windows. (Details at http://www.rlg.org/rtfw.html)

We had already done duospacing for CJK: double-width for data in
the East Asian Character Code (EACC, ANSI/NISO Z39.64) and
single-width for Latin (ASCII plus ANSEL, Z39.47). [Different from
Japanese practice that Frank describes below.]

>Of course most terminals are monospaced, but in Japan they are duospaced:
>ASCII width for Romaji, Katakana, and Hiragana; doublewidth for Kanji.
>As with Roman-alphabet terminals, the host application has a clear and
>unambiguous idea of how characters will line up on the screen, since the
>application and the terminal agree on the width conventions.
>
>Is a similar scheme used for Arabic, ?

The addition that we had to introduce for Arabic was width
distinction by character, rather than at the character set level.

>and if so, is there common agreement
>on which characters are single-width, and which ones are double?

There isn't a standard. It's pretty clear that certain characters
MUST be double-width or single-width for adequate presentation. A
few could go either way, depending on your font design.

>This question relates to Unicode in the design or selection of a Unicode
>font to be used in a multilingual terminal emulator.

The Arabic you get with duospacing is readable, but it isn't
elegant. Your decision on duospacing vs. variable spacing really
depends on how you expect your software to be used.

Various libraries in the US (including the Library of Congress) and
overseas use RLIN to do Arabic script cataloging and for original
script searching.

-- Joan Aliprand
   Senior Analyst
   Research Libraries Group

To: UNICODE@UNICODE.ORG



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:46 EDT