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

Re: Big operators (metafontical answer)

> It is a design feature of the displaystyle operators, that their boxen 
> extend a bit over the baseline. The ``picture'' does not extend over the 
> baseline (except for possibly one pixel because of rounding). I don't know 
> the rationale behind this decision.

The mail from Alan, J"org, and Yannis suggests that my previous
remarks about the `padded' operator were a big fat red herring and
probably totally wrong (not altogether surprising since my impression
of the purpose of the padded operator was formed secondhand by
overhearing some local Metafont conversations). More likely the true
purpose of the padded operator is merely to add the vertical
equivalent of side bearings at the top and bottom of the symbols---so
that Yannis' proposal to draw boxes around the characters and see how
they fit together is after all much to the point. The reason the
padded operator is applied only to the displaystyle symbols is then
clear: they are the only ones that take limits at top and bottom
rather than to the side.

I think this leaves unanswered Justin's original question, which boils
down (I believe) to: why is the height of textstyle sum zero and depth
large, rather than height large and depth zero?  (The third
possibility, height = 1/2 total + math axis, depth = 1/2 total - math
axis, might have been rejected by Knuth in order to avoid building
into cmex a dependency on the cmsy value of math axis. But I am not
sure there is a great deal of value in this, since extension and
symbol fonts probably need to be design-compatible in other respects,
e.g. style of braces).

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