Re: Arabic Requirements

From: Jonathan Rosenne (
Date: Sat Jan 11 1997 - 03:57:12 EST

I am holding a JTC1/SC2/WG2 document of September 1989, "Characters for the
Arabic Script in DP 10646". It contains the requirements for the various
shapes of the Arabic characters and the required ligatures. They are indeed
hand drawn. Its source is ECMA/TC1 Arabic TG and ASMO/TC8 Technical

The current argument on the Arabic ligatures appears very strange to me,
and culturally biased. The Unicode standard defines them as presentation
forms, and specifies their equivalence to their preferred encoding with
basic letters. Thus, they are clearly specified to be equivalent to a
sequence of basic characters, and anyone who can render Arabic correctly
should be able to render them without difficulty.

On the other hand, Unicode and 10646 contain hundreds of pre-composed
Latin, Cyrillic and Greek letters, equally superfluous, equally
decomposable, and this is acceptable because this is what our Western
colleagues are used to.


---------- Forwarded Message ----------


> | The ligatures, especially the Arabic ligatures, were encoded
> | "for compatibility". That is a polite way of saying they were
> | needed to meet some requirement to get the standard approved,
> | or were needed for backwards compatibility to some existing
> | encoding implementation which had a different model of text
> | representation. In the case of the Arabic ligatures, the
> | motivation was entirely for standards approval, because there
> | was no existing implementation.
> What was the specific requirement? I think the Arabic section
> is a mess and I can only imagine that it is the union of several
> fonts. Is that so? what were the fonts?

The requirements were *standards* requirements. They can be traced
back to a series of JTC1/SC2/WG2 resolutions in 1992/1993 and related
input from
national standards bodies, comments on balloting of the standard,
etc. No doubt this is all recoverable from the WG2 minutes and
document archive. However, it was clearly *not* the result of some
existing font. The fonts which were used to print 10646 had to be
specially created for the Arabic ligatures sections, because no one
could locate an existing font which would cover them in sufficient
quality for the printing. The original documents requiring the
ligatures showed them in hand-drawn form. Unicode got the Arabic
ligatures (reluctantly, I might add) from 10646.


--Ken Whistler

> | As the Unicode Standard clearly states, the preferred encoding
> | of Arabic does not use the encoded Arabic ligatures from
> | U+FB50..U+FDFF--and in fact their inclusion in the standard has
> | only made full support of Arabic more complicated, rather than
> | easier.
> And beyond that there are full words and a symbol for "place
> of prayer" that I've never seen anywhere (rather like the
> "hot springs" symbol; perhaps drawn from some guidebook?).

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:33 EDT