At 07:44 PM 11/27/2004, Doug Ewell wrote:
The problem, as Addison pointed out, is that if you use these forms in
text, most searching and sorting operations will fail to recognize them.
That's not the only problem. In some languages other ligatures, such as
fj might be as commonly needed as fi - the set is (intentionally) not
complete and you should not build your text or technology around them.
It is better to use the regular letters and let higher-end software
ligate them as appropriate.
Note that for many languages you need to use ZWNJ to prohibit ligatures
where disallowed by the orthography. Without that information even
fairly high end software cannot correctly ligate these languages.
There are some (in)famous word pairs that are spelled identically,
except for differences in where the ligatures can go. No software
can figure this out - that information must come from the author.
Getting sorting and searching operations to consistently ignore
the ZWNJ is something that has a higher chance of success, compared
to making such software handle long lists of ligatures.
A./
