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

Re: cmex: a draft.

It seems pretty clear, that the NEXTLARGER mechanism for selecting
displaystyle versions of larger operators is just a hack used by Knuth
to avoid having to add \displayfont to the trio of \textfont,
\scriptfont, \scriptscriptfont, with all the extra, mostly useless,
overhead that that would entail. Ideal sizing of large operators, for a
base size of 10pt, would be something like

displaystyle:     14pt
textstyle:        10pt
scriptstyle:       7pt
scriptscriptstyle: 5pt

In practice the symbols found in cmex are used only rarely in
scriptstyle size and 100 times more rarely in scriptscriptstyle size.
The reason for the AMS development of the cmex7 font, however, was to
support the use of \sum, wide accents, and perhaps a few other things
in subscripts and superscripts, which did occur often enough in AMS
publications to make the unsatisfactory nature of using cmex10 for
\scriptfont3 quite clear.

Given that is impractical to change TeX, I think Alan's suggestion to
perpetuate Knuth's hack by keeping two sizes of the large operators in
the cmex encoding, is the best solution for the time being. This means
that for users who choose to load three different sizes of cmex for each
base size, the `large' versions of the operators in the scriptstyle
and scriptscriptstyle fonts will be wasted font positions, as Justin
pointed out. But this seems like a smaller price to pay than the
alternatives.

Michael Downes                              mjd@math.ams.org (Internet)