## new symbols with macros

For feedback and discussion for testers of the MathTime Professional II fonts.

Moderators: PTIForAdmin, WaS, Michael Spivak

Michael Spivak
Posts: 52
Joined: Mon Oct 10, 2005 2:10 pm

### new symbols with macros

Back when we were soliciting new characters for MTProII, some one asked for \barint, saying that his constructed character didn't work well
with limits. Since I was adding lots of characters at the time, I didn't think about this much, and just added \barint, followed later by \slashint.

Some one has now asked for versions of \barint that have the bar near the top or the bottom, for the Riemann upper integral and Riemann lower
integral, which led me to reconsider.

\barint could be defined as follows:

\def\barint{\kern4pt
\raise3.4pt\hbox{\vrule height.8pt width5pt}%
\kern-9pt % -(4pt + 5pt)
\int}

One's first inclination is first set the \int sign, and then add the bar, but
this doesn't give TeX any information about where limits and sub and
superscripts go. The definition above first positions the bar in the right place, then backs up to the initial point, and then prints the \int, so that TeX will now know where things should go. [The \kern4pt, \raise3.4pt,
and the dimensions of the \vrule were just obtained by trial and error.]

Of course, this definition is really only for display style; if one needed it
for both text and display style, one would need to use a \mathchoice (or
a \mathpallete if you're up for it).

Similarly, one might use

\def\Ruint{\kern4.8pt
\raise9pt\hbox{\vrule height.7pt width5pt}\kern-9.8pt\int}

\def\Rlint{\kern2.6pt
\lower4pt\hbox{\vrule height.7pt width5pt}\kern-7.6pt\int}
Michael Spivak
Posts: 52
Joined: Mon Oct 10, 2005 2:10 pm
Actually, it turned out that what was wanted was to have the rules above and below the integral signs, not near the top and bottoms. One might
use

\def\Rlint{\kern1pt\lower10.3pt
\hbox{\vrule height.7pt width4pt}\kern-5pt\int}

\def\Ruint{\kern7pt\raise14.8pt
\hbox{\vrule height.8pt width4pt}\kern-11pt\int}

However, the real concern was apparently that such definitions gave results that look different depending on whether one uses
latex-dvitopdf, latex-dvi-ps-pstopdf, pdflatex, etc. I presume that this
is because these use different methods of determining just which pixels to use for rules, and that differences would only show up on the screen, not