Re: [CSS Text/G.C. for P.M.] Hyphenation & ligatures

On 23 Feb 2009, at 10:41, thomas wrote:

> Hello,
>
> Some fonts have ligatures which are not specified in the Unicode
> standard.  For instance, in 'Linux Libertine', U+E03C (private range)
> is a ligature for 'tt'.
>
> To use the special ligature, one can write:
>
>  fied

This would be a really bad thing to do. Suppose the reader doesn't  
have this particular font, but does have a different font that uses U 
+E03C for something quite different -- such as an arbitrary dingbat or  
a non-Latin character, which would appear as "junk", or even an "ll"  
or "lm" or "bb" ligature, which would completely change the meaning of  
the text.

>
> And to hyphenate the word 'fitted', one can write something like:
>
>  <span style='hyphenate: auto'>fitted</span>
>
> ....according to the 'Generated Content for P.M.' draft.
>
> How to combine both? The idea would be to hyphenate 'fitted' if
> convenient, otherwise, to use the ligature.  The hyphenation model
> offers no way to work with non-standard ligatures.

Font developers should not be encoding ligatures like this in the  
private use area, and authors should not use such codes in text. The  
ligatures should be accessed using "smart font" technologies such as  
OpenType, AAT, and/or Graphite. The text should contain the underlying  
"tt" characters; it is up to the font-rendering mechanism to display  
this using an appropriate ligature, if one is provided.

The same applies to more common ligatures such as "fi", or Arabic  
ligatures like "lam-alef". Don't use "presentation forms" like U+FB01  
or U+FEFB, etc, in text; use the individual characters, and leave  
ligature formation to the fonts.

JK

Received on Monday, 23 February 2009 13:16:21 UTC