add \diagup (U+27CB) and \diagdown (U+27CD) to unicode-math-table

[AlphaZTX]

Status

READY

Description

Add \diagup(U+27CB) and \diagdown(U+27CD) to unicode-math-table.tex according to Section 2.15 of Unicode TR25.

Minimal example demonstrating the new functionality

\documentclass{article}
\usepackage{unicode-math}
\setmathfont{STIX Two Math}
\begin{document}
\[
  \diagup \qquad \diagdown
\]
\end{document}

[davidcarlisle]

Unicode https://www.unicode.org/reports/tr25/ mathclass15.tx has these as class R and MathML Operator dictionary gives them spacing, so if they are added, they should probably be \mathrel not \mathord

That said, most 8bit math font packges make the command \diagup \mathord

amsfonts/amssymb.sty:\DeclareMathSymbol{\diagup}       {\mathord}{AMSb}{"1E}
arev/ams-mdbch.sty:    \DeclareMathSymbol{\diagup}       {\mathord}{mathdesignA}{"1E}
boisik/boisik.sty:\DeclareMathSymbol{\diagup}               {\mathord}{bsksymbols}{"DB}
breqn/msabm.sym:\DeclareFlexSymbol{\diagup}          {Ord}{MSB}{1E}

with MSymbol the exception

mnsymbol/MnSymbol.sty:  \Decl@Mn@Alias\diagup              {MnSyA}{\mathrel}
mnsymbol/MnSymbol.sty:  \Decl@Mn@Alias\ndiagup             {MnSyB}{\mathrel}

several OTF seem to have this in the private use E0A0 slot

concmath-otf/concmath-otf.sty:     \Umathchardef\diagup =         0 0 "E0A0 \relax
erewhon-math/fourier-otf.sty:     \Umathchardef\diagup =         0 0 "E0A0 \relax
xcharter-math/xcharter-otf.sty:     \Umathchardef\diagup =         0 0 "E0A0 \relax

Font coverage is a bit variable

\documentclass{article}
\usepackage{fontspec}
\newfontfamily{\lmm}{Latin Modern Math}
\newfontfamily{\stm}{STIX Two Math}
\newfontfamily{\tgp}{TeX Gyre Pagella Math}
\newfontfamily{\tgt}{TeX Gyre Termes Math}
\newfontfamily{\ncm}{NewCMMath-Regular.otf}
\newfontfamily{\ghm}{GFSNeohellenicMath.otf}
\newfontfamily{\cambriam}{Cambria Math}
\newfontfamily{\xcm}{XCharter-Math.otf}
\newfontfamily{\concm}{Concrete-Math.otf}

\begin{document}
\makeatletter

{\lmm [\char"2751][\char"27CB][\char"E0A0] \f@family\par}
{\stm [\char"2751][\char"27CB][\char"E0A0] \f@family\par}
{\tgp [\char"2751][\char"27CB][\char"E0A0] \f@family\par}
{\tgt [\char"2751][\char"27CB][\char"E0A0] \f@family\par}
{\ncm [\char"2751][\char"27CB][\char"E0A0] \f@family\par}
{\ghm [\char"2751][\char"27CB][\char"E0A0] \f@family\par}
{\cambriam [\char"2751][\char"27CB][\char"E0A0] \f@family\par}
{\xcm [\char"2751][\char"27CB][\char"E0A0] \f@family\par}
{\concm [\char"2751][\char"27CB][\char"E0A0] \f@family\par}


\end{document}

But I suppose adding here might encourage fonts to converge, but mathord or mathrel? amssymb compatbilty means I think the ame \diagup should be mathord, but I wonder if a mathrel version should be supplied as well...

[AlphaZTX]

Maybe we can try the following code:

\UnicodeMathSymbol{"027CB}{\diagup                   }{\mathord}{mathematical rising diagnal}%
\UnicodeMathSymbol{"027CB}{\diaguprel                }{\mathrel}{mathematical rising diagnal (relation)}%

U+27CD is similar.

[davidcarlisle]

something like that yes (\mathrel for the second one of course:-)

Main argument against adding these is that latin modern math doesn't seem to have anything usable, so the default behaviour of the new commands would be a missing glyph warning, which is a bit unfortunate

[wspr]

I'm really surprised these were missing given \diagup is in amsmath. Based on a very cursory scan of how \diagup is used in the wild, I think \mathrel doesn't seem right:

https://arxiv.org/search/advanced?advanced=1&terms-0-operator=AND&terms-0-term=%5Cdiagup&terms-0-field=all&classification-physics_archives=all&classification-include_cross_list=include&date-filter_by=all_dates&date-year=&date-from_date=&date-to_date=&date-date_type=submitted_date&abstracts=show&size=50&order=-announced_date_first

I'm inclined to only provide \mathord for now.

[davidcarlisle]

I'm inclined to only provide \mathord for now.

Given amssymb makes them mathord I guess that is best. Bit of pain that mathml4 and mathml-core give them a default 3mu space left and right following the R(elation) classification from Unicode TR25 but getting all these things lined up is hard.

[davidcarlisle]

see comments already posted

[wspr]

Many thanks!