[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ligatures in math font.

 
>if anyone loads an
>>MC-encoded font containing upright glyphs, in order to get the
>>effect of ${\bf waffle + \Gamma}$.
>
>One of the major thing we decided is that MC IS NOT FOR MULTILETTER
>identifiers.   So this example is no good.
 
Another one of the goals is to be compatible with current plain TeX.  At 
the moment, the input $\bf waffle + fit$ is acceptable input for 
$\mathbf{waffle} + \mathbf{fit}$.  The TeXbook (p. 164 in the 1991 
printing) states that the input:
 
   $\bf a+b=\Phi_m$
 
will produce the `a', `b', `\Phi' and `m' in bold, and the `+' and `=' in 
non-bold.  If we want to be upwardly compatible with this, we have to 
allow for an MC-encoded font containing bold upright Latin glyphs, *and* 
the five ligatures <fi>, <ff>, <fl>, <ffi> and <ffl>.
 
If we don't allow these ligatures, then there are documents which 
currently work (such as `Let $\bf Aff$ be the category of Affine logics' 
or `${\bf if}\; x<0\; {\bf then}\; x:={-x}\; {\bf fi}$') which will break 
with the new encodings.
 
(Yes, I know it would be `better' to input these as `Let 
$\mathbf{Aff}$...' or `$\mathbf{if}...' but we don't have the option of 
rewriting the TeXbook.)
 
This argument also applies to \sf, \it, etc.  For example, input such
as `$\sf \Omega = fix\;I$' is very common in papers on the 
lambda-calculus.  
 
There is a conflict between two of our goals: upward compatibility is 
incompatible with insisting that MC has no text ligatures.  So the 
question is, which of these is more important?
 
IMHO, upward compatibility is much more important than ligaturing.  I'd 
be very worried that if we produce an encoding where 1% of `kosher' plain 
TeX documents break, then we'll have a great deal of difficulty 
persuading users to switch to the new fonts.  I don't particularly like 
including text ligatures in math fonts, but I think it is necessary.
 
Alan.