[XeTeX] Single glyphs from a font

[Ross Moore]

Hi Joel,

On 09/06/2009, at 12:11 PM, Joel C. Salomon wrote:

> Ross Moore wrote:
>> \makeatletter
>>     \let\UnicodeMathSymbol\um at mathsymbol@noparse
>>     \UnicodeMathSymbol{"02026}{\mathellipsis}{\mathrel}{ellipsis}%
>>     \UnicodeMathSymbol{"02032}{\prime}{\mathord}{prime}%
>
> OK, we’re filling what’s missing from unicode-math.tex.  I tried  
> that,
> but I missed the
>     \let\UnicodeMathSymbol\um at mathsymbol@noparse
> that makes the command work!

Yes, indeed.

>>     \let\prime ′
>>     \let\mathellipsis …
>
> Noting for the record that defining \mathellipsis causes amsmath’s
> \dotsc &c. to use that symbol.  (You know that, but I only found that
> out by searching my MiKTeX installation tree.)

Use TeX's internal tracing, and study the resulting output
--- either in a Console/Log window, or view the .log  file.

e.g.
  \[ x_0, x_1,  \ldots, x_n, {\tracingall\dotsc} \]
should reveal how \@ldots  arises.

\[ x_0 + x_1 + \cdots + x_n, {\tracingall\dotsb} \]
should reveal how \@cdots  arises.

Do these without the  \let\@....  commands above,
or by using expandable macros instead; viz.
   \def\@ldots{…}
   \def\@cdots{⋯}

>
>>     \let\@ldots …
>>     \let\@cdots ⋯
>> \makeatother
>
> What are these definitions for?  These commands are about to be
> unaccessible after \makeatother.  Is there some package that’s  
> expecting
> these names?

They are part of the expansion, as defined in  amsmath.sty
of the user-level commands  \ldots  and  \cdots .
  \tracingall  is a good way to find out what is happening
to macros, without trying to read the coding inside packages.
Beware though, it can produce *lots* of output, even when
you use braces {\tracingall ... } to limit the scope.

>
> (B.T.W., unicode-math.tex already declares ‘⋯’ to be a  
> \mathrel, so I
> omitted that line.)
   Fair enough. No need to repeat it.

>
> So basically, the reason \tag{\ref*{abc}$′$} didn’t work before  
> was that
> the TeX didn’t know what math category the prime character fell  
> into, so
> it got ignored?

Yes, sort of.
Tracing with  \tag{$\ref{abc}\tracingall′$}
gives:

>> {into \tracingonline=1}
>> {the character ′}
>> {math shift character $}
>> Missing character: There is no ′ in font cmr10!


So unless a character has been declared for math by unicode-math,
it will have its default meaning from non-XeTeX LaTeX.
But there is no such character in those old-style fonts, so it
just gets left out, and takes up no typesetting space.


>
>> Now even this works:    \tag{$\ref{abc}'$}
>> using the ordinary ASCII single-quote/apostrophe.
> Not quite; $'$ has some superscripting magic associated with it.  Try
> $\prime{}'$ and you’ll see the difference.  (Without the ‘{}’,  
> you’ll
> get a weird primed prime.)

That is TeX's  \mathcode"8000  trick, which allows ordinary
letters to become macros when in math-mode.
Use \tracingall to see what happens.

>> '->^\bgroup \prim at s
>> {superscript character ^}
>> {entering math group (level 6) at line 40}
>>
>> \prim at s ->\prime \futurelet \@let at token \pr at m@s
>> {the character ′}
>> {\futurelet}
>> {changing \@let at token=the character *}
>> {into \@let at token=blank space  }
>>
>> \pr at m@s ->\ifx '\@let at token \expandafter \pr@@@s \else \ifx ^ 
>> \@let at token \expandafter \expandafter \expandafter \pr@@@t \else  
>> \egroup \fi \fi


It is "looking ahead" to see if the next character is another prime,
or the ^ symbol.
This makes it very convenient to do  f'' and f'''  etc.
But other complicated superscripting should not be done this way.


>
> (B.T.W., re. the ‘*’:  I’m using hyperref; Ignore it.)

Aaah; OK. I should have guessed this.


>
> Thanks for the help,
> —Joel Salomon


Hope this helps,

	Ross

------------------------------------------------------------------------
Ross Moore                                       ross at maths.mq.edu.au
Mathematics Department                           office: E7A-419
Macquarie University                             tel: +61 (0)2 9850 8955
Sydney, Australia  2109                          fax: +61 (0)2 9850 8114
------------------------------------------------------------------------