[XeTeX] Minor problem with kerning in Ezra SIL (Hebrew)

[Jonathan Kew]

On 18 Jan 2005, at 8:57 pm, Robert Voogdgeert wrote:

>> Note that these two sequences are defined to be canonically equivalent
>> (according to the Unicode standard), and therefore should be
> completely
>> interchangeable; however, Ezra SIL will only render the second version
>> properly.
> Jonathan,
>
>
> Thanks for your response. I use to copy and past Hebrew texts from
> Accordance, so I wasn't aware of the order. Also, I never read the FAQ,
> which appearantly I should have... By the way, I found out that the 
> same
> problem holds for any other vowel sign or cantilation mark in
> combination with dagesh, e.g. consonant > chireq > sheva > munach 
> should
> be changed in consonant > sheva > chireq > munach.
>
> To my surprise, whatever order is used in TextEdit, they both give the
> wrong kerning, as against XeTeX, that indeed give the correct kerning 
> if
> the dagesh immediately follows the consonant.

That's to be expected; TextEdit doesn't handle the OpenType features of 
the font. Very few apps on OS X support OpenType layout behaviors at 
this point; one that probably would work (though I haven't tested this) 
is Mellel.

> I will try and see whether I can get things right with find/replace in
> BBEdit, though this will be quite a delicate process, because there are
> many possible combinations with dagesh.

True. As a general rule (check the font FAQ!), I think the dagesh 
should come immediately after the consonant, before other diacritics; a 
suitable regular-expression replace operation could probably do this.

JK