consensus Call: TATWEEL

Kenneth Whistler kenw at
Wed Mar 25 22:56:26 CET 2009

> If, in our choices of allowable code points, we constrain Arabic  
> (and Farsi and ...) to choices which look correct to Latin  
> expectations, non-descending rather than descending, inaesthetically  
> dense rather than combining and extended, to "2nd gen" computer Arabic  
> (and ...) -- then we may be "solving" one problem by creating another.

I think this has nothing whatsoever to do with the correct
choice of allowable code points for Arabic. You are talking
about typographical layout concerns for Arabic, which are
issues of rendering systems and fonts -- not the choice of
the basic letters of the Arabic script appropriate for spelling
out domain name labels.

> ... but we do have the choice to ask if the people who create  
> the basic tools for writing Arabic (and ...) need what we are asked to  
> ban, and because of our shared Latin (and other scripts lacking one or  
> both of the properties, descent and vertical and horizontal  
> interaction) limitations, simply fail to appreciate.

Not at all, if you are talking about U+0640 ARABIC TATWEEL, in

Elimination of U+0640 as PVALID for domain names is not the
same as claiming that kashida justification for Arabic text
would be disallowed, even for domain names.

U+0640 ARABIC TATWEEL is an old *hack* for enabling very simplistic
rendering systems to do (very) ugly kashida justification when
constrained to fixed-width character cell graphics. Sophisticated
modern Arabic rendering systems do not drop in a *character*
code for kashida justification, any more than sophisticated
modern Latin rendering systems require dropping in thin space
characters between letters in order to do intercharacter
space justification. U+0640 ARABIC TATWEEL is simply not
needed at all in domain name labels, as attested by numerous
Arabic script-using participants in this working group already.


> If you really do think this is "bold" and "italic" you simply aren't  
> getting it.
> Eric

More information about the Idna-update mailing list