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

rsfs slant problems



Barbara Beeton has circulated some recommendations for cursive fonts

     I will call Ralph Smith's Formal Script a "cursive" font to
avoid confusion with "script" as in \scriptstyle meaning "index or
exponent style".

     I am hoping to see Ralph Smith's rsfs enhanced and made
available in Type1 format.

     Barbara Beeton has circulated some desiderata for cursive
fonts that include:

 > composition constraints:
 > 
 >  - assume basic text is 10pt Times Roman or Century Schoolbook
 > 
 >  - hairlines must be thick enough to prevent breakup in sub- and
 >    superscripts, and in reprints from first printing; first-order
 >    script size is approximately 70% of basic text, second-order
 >    size, approximately 50%
 > 
 >  - although rare, some bold script letters are sometimes requested;
 >    normal weight must be distinguishable from bold
 > 
 >  - slope of letters should not be extreme, in order to minimize
 >    problems in applying accents and sub/superscripts; err toward
 >    the upright rather than exaggerated slope
 > 
 >  - x-height should be closer in size to basic text font than is
 >    usual; traditional script fonts tend to have rather low x-height
 > 
 >  - this font need not be usable for continuous text

Let us discuss some of these.

 >  - hairlines must be thick enough to prevent breakup in sub- and
 >    superscripts, and in reprints from first printing; 

OK.  But cursive in *2nd Order* sub/super positions is a typographical 
no-no.  Such problems are as well left unsolved.

 >    first-order
 >    script size is approximately 70% of basic text, second-order
 >    size, approximately 50%

Hopefully no. Because of metaness, script *height* is only one of several
"size" parameters that are varying.  *Height* is a constraint; the 
other paramaters are free to solve the typographical problems that the 
constrained height brings.

 - although rare, some bold cursive letters are sometimes requested;
   normal weight must be distinguishable from bold

Well, there is only one sort of bold that should be provided,
namely a bold for mathematics in bold titles.  If you accept this,
then it is *untrue* that it must be easily distinguished from 
ordinary cursive.  The constraint is that it look in balance with
cmbx10--12.

 - slope of letters should not be extreme, in order to minimize
   problems in applying accents and sub/superscripts; err toward
   the upright rather than exaggerated slope

The advice "err toward the upright rather than exaggerated slope"
does reduce some problems, notably those that arise when one places
a cursive letter in a super/sub position. 

     Maybe I am missing something basic, but I fail to see why
slope causes serious problems with respect to any of the following:

     --- applying math accents to cursive characters
     --- applying (non-cursive) sub/super-scripts to cursive characters

the reason being that TeX has adequate parameters to get the right
placements - respectively:

     --- the implicit kerns with the "skew character" of the cursive font
     --- the italic corrections in the cursive font.
   

     As for the problems that arise when one places a slanted letter in a
super/sub position, I am afraid TeX has nothing to offer much better than
"hand setting". 

     Why? My impression is that Knuth avoided this problem by special boxy
ad hoc design of certain letters such as y and j in cmmi.  I would
recommend the same policy for rsfs.  On the other hand, Knuth's "skew
character" trickery could I believe be extended to solve these problems in
some future e-TeX.  This is a hint for Mathias Classen and friends.

                     Cheers

                            Larry Siebenmann


PS.  Has someone redesigned MathTime to avoid the problem last discussed?
I am unlikely to use  MathTime again for heavy math until that happens.
Once burned twice shy.