Re: your huge STIX table
Ross Moore
On 22/11/2006, at 1:38 AM, Barbara Beeton wrote:
> hi, ross,
> wow, you really have looked through the table
> quite thoroughly. i'm just back from the tug
> meeting, and have a bunch of things piled up,
> so i can't answer in detail right now, but i
> can address a few things quickly.
>
> I've found your huge table, at:
> http://www.ams.org/STIX/bnb/stix-tbl.ascii-2006-10-20
> and it's explanation at:
> http://www.ams.org/STIX/bnb/stix-tbl.layout-2006-05-15
>
> there are some additions, not yet posted on line,
Does this include the following ?
U+027CA VERTICAL BAR WITH HORIZ. STROKE
U+02B00 --> U+02B11 (Misc. Symbols & Arrows Block)
especially U+02B0E, U+02B0F, U+02B10, U+02B11 which
are similar to glyphs in Lucida New Math Arrows font.
> and a few glitches reported. i'll let you know
> when the new version is up.
>
> All very impressive indeed, and exciting for the future.
>
> thanks.
>
> Firstly, column 57 is a letter denoting math-class.
> What is to be inferred when this is empty ?
> Does it mean that you don't expect this character to ever be used
> within mathematics ? Or could it, in some cases, indicate that its
> not obvious what class to assign, or could it just be an
> oversight ?
>
> any of these, depending, but i think mostly, the
> class wasn't immediately obvious to me. your
> suggestions will be very helpful.
OK. I may spend some time looking at these.
>
> Or would the meanings of such characters be done in another way,
> within mathematics; e.g. 000BD onehalf ?
>
> recommendation for this one, at least, is that
> it be done differently in math, so that the style
> matches, e.g., \frac{1}{n} .
With MS supporting math via Cambria, and when the STIX fonts
finally appear, then it will become common for people
to Copy/Paste mathematical expressions between documents
constructed using different applications.
(Also, I'm doing some work that will allow this from LaTeX docs
using CM and MSAM, MSBM fonts.)
Thus we will find characters such as U+000BD appearing within
the source text for (La)TeX documents.
In my opinion, these should be supported directly, since the
meaning is clear.
The argument for "style matches" is quite sound, so my proposal
would be to use the mathcode-8000 trick;
e.g., with TeX coding such as:
\mathcode"BD = "8000
{\catcode"BD \active
\gdef^^bd{\frac12} }
This way you get the best of both worlds.
For example, this works perfectly:
\documentclass{article}
\mathcode"BD = "8000
{\catcode "BD \active
\gdef^^bd{\frac12} }
\begin{document}
$\show^^bd ^^bd$
\end{document}
even though the .log file shows:
> the character Ω.
l.6 $\showΩ
^^bd$
?
Of course for other fractions in the range: U+02153 --> U+0215E
we need an extension of TeX, such as XeTeX, that can handle direct
input of higher codes.
e.g.
\documentclass{article}
\mathcode"BD = "8000
\mathcode "2153 = "8000
{\catcode "00BD \active
\catcode "2153 \active
\gdef^^^^00bd{\frac12}\gdef^^^^2153{\frac13}
}
\begin{document}
$\show^^^^00bd ^^^^00bd \tracingall^^^^2153$
\end{document}
works fine with xelatex , producing .log entries:
> the character ½.
l.10 $\show½
^^^^00bd \tracingall^^^^2153$^^M^^Mar} @}
?
{the character ⅓}
⅓->\frac 13
\frac #1#2->{\begingroup #1\endgroup \over #2}
#1<-1
#2<-3
{begin-group character {}
{\begingroup}
{the character 1}
{\endgroup}
{\over}
{the character 3}
{end-group character }}
{math shift character $}
[1] (./test.aux)
)
Output written on test.pdf (1 page).
This is the kind of thing that I propose for characters in the special
math-families, such as \mathrm \mathbb \mathfrak etc.
Unfortunately I cannot give explicit examples, as XeTeX
doesn't yet support direct entry of Plane-1 codes.
>
> review of your examples to come later.
> 003AB --> 003B0 (small Greeks with accents -- would be done
> differently?)
> 003CA --> 003CE ( ditto )
>
> ... are now normally done differently in TeX, this is
> something that
> could easily change, especially with XeTeX emerging onto the
> scene.
> It will become possible to copy/paste strings with such characters
> into LaTeX source.
> They should then typeset "correctly", both in math and in text-
> mode.
> Thus I think there should be TeX macro-names assigned, and a
> math-class
> wherever appropriate.
>
> no. the consensus opinion, when i checked with
> both the unicode and mathml folks, is that accented
> characters, with very few exceptions, should be used
> only in text, not math. so you will find that in the
> unicode tables, the math attribute is assigned only
> to "bare" characters.
Fine --- in principle.
However, in practice things are a little different.
For example, using Mathematica, the characters for \"a and \ddot{a}
appear to be identical. It is very easy (for physicists in particular)
to mix these up.
The solution is similar to above. Set the \mathcode to "8000
for accented Latin letters, such as \"{a} --> U+000E4 ,
so that they become active when entered in math-mode.
The resulting macro-expansion then can be set to use the
math-accent instead.
e.g. ^^^^00e4 ---> \ddot{a} (when in math-mode).
> the exceptions that i can think
> of appear only in the letter-like symbols block (in
> the 20xx or 21xx range somewhere).
>
> Similar comments apply for the following:
>
> 02060 --> 02063 (various amounts of space )
> 02070 --> 0208E (sup/sub-scripted letters and punctuation )
> 02153 --> 0215E (small number fractions , usually done with
> \frac )
> 022EE --> 022F1 (dots, at different angles)
> 02320 --> 02321 (top/bottom half of integral )
> 02329 --> 0232A ( angle brackets )
>
> some of these certainly deserve a second look, but
> not the sub/sup letters or fractions.
Same practical considerations apply for these too.
> remaining examples will be checked too.
>
> most of the box-drawing pieces
>
> are these really math? don't remember offhand who
> requested these, but it wasn't obvious that they
> would be used for anything but graphics on a screen.
> if you have support for "math" use, i'd like to hear
> the reasons.
> thanks for the detailed comments. more later.
> cheers.
> -- bb
All the best.
Merry Christmas and Happy Holidays,
Ross
------------------------------------------------------------------------
Ross Moore ro...@maths.mq.edu.au
Mathematics Department office: E7A-419
Macquarie University tel: +61 +2 9850 8955
Sydney, Australia 2109 fax: +61 +2 9850 8114
------------------------------------------------------------------------