Re: Unicode and math symbols

["Martin J. Duerst" <mduerst@ifi.unizh.ch>]

On Sun, 2 Mar 1997, Berthold K.P. Horn wrote:

> 
>    > And yes, just about all 30,000 fonts in Type 1 format have fi and fl.
> 
>    Nothing bad about that. Any good font should have these ligatures,
>    and a few more. But it shouldn't matter which number they have,
>    you should be able to send "f","i" to the font and it should
>    do the substitution on its own.
> 
> That I do not agree with.  Yes, low end applications will work this way
> with GX and OpenType.  But high end applications will need to work at
> this level themselves, *not* let the system machinery do it.  And after
> the application decides to use a ligatures it must have a way of
> selecting it from the font.

I agree that "take this character string and typeset it in a nice way"
is not the only method a font should provide. But it is a very important
one, especially for low end applications.


> If TeX decides after much cogitation that it wants to use ffi it has to
> have a way to select that from the font.

For an application, there can be other ways to select ligatures. One is
to insert ZWNJ (zero width non-joiner) or similar codepoints at positions
where ligaturing has to be suppressed. The other would be to use a method
of the font that allows the application to ask what ligatures are present.
I.e. calling this method with "ffi" will return true for fonts that
contain an "ffi" ligature and false for fonts that don't. After that,
the application could decide on its own whether e.g. it wants to try
to use an "ff" ligature or an "fi" ligature, or none at all,...


Regards,	Martin.