[XeTeX] Correct LaTeX-logo – how?

[Bruno Voisin]

Le 7 mai 06 à 16:24, Jonathan Kew a écrit :

> I guess this is a question for the LaTeX experts.... what's the best
> way to overcome Babel's redefinition of the logos? It's probably
> happening because Babel tries to ensure that the logos are set with a
> certain font encoding (in case you're using Russian or something for
> your main text), but this means that it defeats fontspec's use of the
> "U" encoding for Unicode fonts.

This is how I had modified the LaTeX logo a few months ago, to be  
used in a document typeset in Optima (which doesn't include the small  
caps font used for the A in the standard logo). Babel was also used  
in this document. Together with graphics.

\makeatletter
\renewcommand{\TeX}{\org at TeX}
\renewcommand{\LaTeX}{\org at LaTeX}
\makeatother
\newsavebox{\LTXT}\newlength{\LTXTht}
\newsavebox{\LTXA}\newlength{\LTXAht}
\DeclareRobustCommand{\LaTeX}
   {L%
    \kern-.30em%
    \savebox{\LTXT}{T}\settoheight{\LTXTht}{\usebox{\LTXT}}%
    \savebox{\LTXA}{\scalebox{.8}{A}}\settoheight{\LTXAht}{\usebox 
{\LTXA}}%
    \raisebox{\LTXTht}{\raisebox{-\LTXAht}{\usebox{\LTXA}}}%
    \kern-.15em%
    \TeX}

The height of the T and of the A, scaled by .8 for the latter, were  
measured and used to move the A vertically such that its top was (in  
theory) aligned with the top of the T. This worked reasonably well,  
but not perfectly, in both upright and italic shape:

-------------- next part --------------
A non-text attachment was scrubbed...
Name: Image 3.png
Type: image/png
Size: 4233 bytes
Desc: not available
Url : http://tug.org/pipermail/xetex/attachments/20060508/e37ca04d/attachment-0002.png 
-------------- next part --------------
  
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Image 2.png
Type: image/png
Size: 4398 bytes
Desc: not available
Url : http://tug.org/pipermail/xetex/attachments/20060508/e37ca04d/attachment-0003.png 
-------------- next part --------------

With the new version of XeTeX setting glyph metrics, this might  
become better. I'll look into compiling 0.991 when (if) I have time.

Bruno Voisin