[XeTeX] t caron problem (gentium font?)

[Ross Moore]

Hi Arthur, Jonathan, Václav,

On 19/04/2009, at 8:17 AM, Arthur Reutenauer wrote:

>>                                                          I guess you
>> used the sequence <t, combining caron> in your text

With  xunicode , you can also use  \v{t}  or  \tcaron .
This will place the appropriate  \char"0165 .

At present  xunicode  does no checking of whether this character
actually exists within the font you are trying to use.

If you also include the line

  \UndeclareUTFcomposite[\UTFencname]{x0165}{\v}{t}

then  \v{t}  will give you  t\char"030C  using the
"Combining caron" at U+030C .
But again there is no checking that the font actually has
this character, nor whether it composes correctly.

This is the 2nd time in recent weeks that there has been
a problem with a font not supporting an accented character
   (recall \hungo : \H{o} ).


My question is whether or not it is worthwhile building
in checks of the font for the existence of the glyph?
e.g.,
programming the logic
   a.  if the precombined character exists, use it;
       otherwise
   b.   if the combining accent exists, use that combination;
       else
   c.   if a non-combining accent exists, use LaTeX's command
         \set at accent  which positions the pieces in boxes.

In favour of this, you'll always get the best that the font
can support, ...

  ... though in this case  \set at accent  will give something that
looks wrong, whereas it looked correct with \hungo .


On the downside,

   1.  you won't know in advance what you will get, when just
       using a font that you have not previously checked for
       the availability of the characters you need;

   2.  it'll be harder to detect insufficient support for some
       characters by just visual checking of the PDF produced;

   3.  "missing character" messages will no longer appear
       in the .log  file;

   4.  extra processing will be done every time an accented
       character is requested --- perhaps even for standard ones.

Besides,  xunicode  already has the ability for customised
encodings, using the optional parameter:
e.g.
    DeclareUTFcomposite[<encoding>]{x0165}{\v}{t}

That <encoding> parameter is available to declare customised
encodings for fonts that support restricted sets of characters.
Someone XeTeX- and font-aware can write a package specific to
the font, allocating a name for its encoding and declaring
all the characters it supports, and undeclaring others that
need to be constructed using other methods.

This requires a fair bit of initial work, but eliminates
all the downsides of continual checking.

>
>   You're most certainly right, Jonathan.  How could I, of all people,
> miss that!

You're forgiven.  :-)

But how should we move ahead, coping best with the proliferation
of inadequately-constructed fonts that some people will inevitably
choose to use ?


>
> 	Arthur


Cheers,

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