[XeTeX] something wrong with xunicode.sty [WAS: do you see...]

[Ross Moore]

Hi Ulrike, Francois, and others,

On 16/06/2008, at 3:47 AM, Ulrike Fischer wrote:
> Am Sun, 15 Jun 2008 19:11:13 +0200 schrieb
> francois.patte at math-info.univ-paris5.fr:
>
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> Bonsoir,
>>
>> According to the answer by Syros Apostolos to my previous mail,
>> character "\b n" (n , b , d whatever) is undefined if you load
>> xunicode.sty, and is defined if you don't load it....
>>
>> I don't understand why because this can be read in xunicode.sty:
>>
>> \DeclareUTFcomposite[\UTFencname]{x1E47}{\d}{n}
>>
>> Who can help?
>
> The \b is defined in xunicode as
>
> \DeclareEncodedCompositeCharacter{\UTFencname}{\b}{0332}{005F}

No; this isn't a *definition* of \b .
It is a *declaration* of what to do, under certain circumstances.

In the case at hand, it is actually irrelevant, since the
appropriate circumstances do not arise.

These later declarations take precedence:

  \DeclareUTFcomposite[\UTFencname]{x1E48}{\b}{N}
  \DeclareUTFcomposite[\UTFencname]{x1E49}{\b}{n}

so that with the combination of accent \b  and letter  n
XeTeX is asked to use  \char"1E49 .

>
> There 0332 refers to an "combining low line"-char and 005F is the low
> line.

If you want to test availability of the "combining low line" accent
character, with the letter 'n' you can do it using:

   \b{{{n}}}

Note the need for at least 3 levels of bracing, as each of
  \b n , \b{n}  and  \b{{n}}  all result in  \char"1E49
being used, rather than   n\char"005F  .


>
> This works fine e.g. with the lmodern fonts, but fails with your
> Linotype because it seems not to have the 0332-char. You should always
> expect that some chars doesn't exist in some fonts and so some
> definitions can fails.
>
> If you don't load xunicode \b will fall back to its default
> OT1-definition (I didn't try if it works).

This will work, but visually only.
If you try to copy/paste from the resulting PDF then you
will not get anything sensible for the accent character.

In fact, using Preview as the PDF browser the accent does not
copy at all, whereas with Adobe Reader you get a macron accent.

Whereas, with xunicode.sty loaded, you'll get the correctly
accented letter, useable in other UTF8-aware applications,
on any computing platform (Mac OS, Linux, Windows, etc.).


>
> -- 
> Ulrike Fischer


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
------------------------------------------------------------------------