[XeTeX] XeTeX 0.92 available

Jonathan Kew jonathan_kew at sil.org
Thu Mar 10 11:27:53 CET 2005


On 10 Mar 2005, at 10:07 am, Adam Lindsay wrote:

> Otared Kavian said this at Thu, 10 Mar 2005 08:57:49 +0100:
>
>> they correspond to the "period" obtained via an "Arabic" keyboard,
>> that is they are nonroman punctuation marks.
>
> Did you (or I) mis-type? I was asserting that they are *roman*
> punctuation marks: U+002e '.'
>
> It's not implausible that a given Arabic keyboard layout is spitting 
> out
> roman punctuation. I was asking what happens when you insert the 
> "proper"
> Arabic full stop.

By which you mean U+06D4, I assume. That may well behave differently, 
as it has different Unicode bidi properties ("AL" rather than "ON", 
IIRC).

However, it is entirely legitimate to use U+002E as a period within 
Arabic text. This is not a *roman* punctuation mark, it is a *generic* 
punctuation mark. (Note that, like others, it has "neutral" 
directionality, and its script classification will be something like 
"common" or "inherited"--don't have the files at my fingertips.)

U+06D4 is rarely used, except in certain styles of Arabic script (in 
particular, Nastaliq); IMO, it shouldn't really have been encoded at 
all, it should have been considered a stylistic variant of the normal 
period. The form shown for this character in the Unicode charts is 
characteristic of Nastaliq style script, but is not used in the 
Naskh-based style that is the basis for most text fonts.

Note also that only a few punctuation marks are provided with 
right-to-left counterparts in the Arabic block (basically, those that 
have substantially different shapes: comma and semicolon). Others are 
assumed to be used from the "Basic Latin" (a misnomer, given that it 
includes non-script-specific characters!) and "General Punctuation" 
blocks as needed. So if the problem arises because of the use of a 
"neutral" rather than specifically R-L punctuation character, then 
there are others besides period that could trigger it (colon, dashes, 
quotes, etc.), and we need to find and solve the root issue, whether in 
the XeTeX engine itself or the ConTeXt package.

JK



More information about the XeTeX mailing list