Re: Latin ligatures and Unicode

From: peter_constable@sil.org
Date: Mon Dec 20 1999 - 09:32:56 EST


       ME>You misunderstand the ZWL. What you've described is an
       inputting operation. If German is well-behaved enough to trust
       morphological analysis for ligature behaviour, then the ***ZWL
       can be inserted automatically by software***. This would work
       as well for fi, ffi, ffl &c ligatures. Unpredictable, rare and
       nonce ligatures can be handled by the user inserting the ZWL
       manually (as we do the SOFT HYPHEN).

       [emphasis added]

       JJ>In this case, it's possible to specify via either OpenType
       or Apple Advanced >Typography (AAT) tables in a TrueType font
       that such ligatures are required >when using your font.
       Applications which take advantage of these >technologies will
       do the right thing. (For the record, you can also specify
>required ligatures in Metafonts for use with TeX.)

       ME>No, it isn't. Because it is impossible to easily apply
       ligation in the many different manifestations which are
       evidenced. In my paper on ZWL I have described the use of
       Gaelic ligatures in printed books; it is NOT possible to decide
       on two or three global settings which will give the correct
       behaviour when turned on or off. The REALITIES are that what we
       need to represent, in Latin (Roman, Gaelic, and Fraktur),
       Greek, Old Church Cyrillic, Armenian, Runic, Etruscan, and so
       on are far more complex than the fun one can have with Hoefler
       Text.

       If software can automatically do something in one place, it
       seems to me that it should also be able to do an equivalent
       thing in another place. I.e. if software can automatically add
       ZWL at appropriate places to create the desired ligature glyphs
       on output, then software should also be able to automatically
       select the desired sequences of character codes (sans ZWL) to
       substitute by ligature glyphs. At least, that seems to make
       sense to me unless there's more to this story than I'm not
       aware of.

       For unpredictable ligatures, it should be possible (i.e. 'it is
       technologically feasible', not 'software ought to be able') to
       apply style information to the relevant run of text enabling a
       feature that causes ligation.

       Peter



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:56 EDT