[XeTeX] Proper mapping for ? in bibliography - {\'\i} or {\'i} ?

[Bruno Voisin]

Le 19 juin 07 à 21:51, Nicholas Riley a écrit :

> I was attempting to fix the typesetting of a í in an author's name in
> a BibTeX file which BibDesk encoded as {\'\i}; instead of appearing
> as desired, I got a dotless i followed by a square.
>
> After poking around xunicode.sty a bit I added:
>
> \DeclareUTFcomposite[\UTFencname]{x00ED}{\'}{\i}
>
> which seems to work, but I am wondering if this is the "right" thing
> to do or if BibDesk should be generating {\'i} instead...or something
> else?

Le 20 juin 07 à 05:57, Nicholas Riley a écrit :

> xunicode works fine with BibTeX, though I don't currently have it
> configured for BibDesk previews because XeTeX is so slow to render
> (and my machine is old).  xunicode just, (seemingly) omits a
> definition for {\'\i} which the line I provided remedies.

Now I understand. I hadn't suspected the problem could lie within  
xunicode.sty.

That's odd. The version history for xunicode.sty does mention dotless i:

% v0.3  2004/09/10
%       support for accented \i and \j as well as plain i and j letters.

But on one side you've got:

\DeclareUTFcomposite[\UTFencname]{x00EC}{\`}{i}
\DeclareUTFcomposite[\UTFencname]{x00ED}{\'}{i}
\DeclareUTFcomposite[\UTFencname]{x00EE}{\^}{i}
\DeclareUTFcomposite[\UTFencname]{x00EF}{\"}{i}

and on the other side there's:

\DeclareUTFcomposite[\UTFencname]{x0128}{\~}{I}
\DeclareUTFcomposite[\UTFencname]{x0129}{\~}{i}
\DeclareUTFcomposite[\UTFencname]{x0129}{\~}{\i}
\DeclareUTFcomposite[\UTFencname]{x012A}{\=}{I}
\DeclareUTFcomposite[\UTFencname]{x012B}{\=}{i}
\DeclareUTFcomposite[\UTFencname]{x012B}{\=}{\i}
\DeclareUTFcomposite[\UTFencname]{x012C}{\u}{I}
\DeclareUTFcomposite[\UTFencname]{x012D}{\u}{i}
\DeclareUTFcomposite[\UTFencname]{x012D}{\u}{\i}
\DeclareUTFcomposite[\UTFencname]{x012E}{\k}{I}
\DeclareUTFcomposite[\UTFencname]{x012F}{\k}{i}
\DeclareUTFcomposite[\UTFencname]{x0130}{\.}{I}
\DeclareUTFcharacter[\UTFencname]{x0131}{\i}

So it seems care was taken to make i and \i equivalent in some cases  
(like \~i and \~\i) but not others (like \'i).

I guess this one is for Ross Moore, the author of xunicode.sty.

Bruno Voisin