Andreas,

> >   The assumption behind it is that the width of a "character" stays
> > same wherever it is.  Some characters in some languages change its
> > "width" based on the characters around it.
> 
> Almost all languages/fonts have glyphs which should be spaced differently
> based upon context. It's typically represented by kerning pairs where
> character pairs get "extra spacing" values.

  You're right.  Even the NewYork fonts and other StrikeFonts in
default Squeak are designed to do the a simplified job, it is still
possible that the "width" calculated as the sum of the content
character is not equal to the actual string.

> Often, this is just a crude
> workaround for ligatures (such as ff) which really should be represented by
> an individual glyph but it's "better than nothing".

  Yup.

  I have an impression that OpenType utilize the "compound glyph"
mechanism, which the TT renderer in Squeak can handle, to do the
"ligatures" in Thai, Devanagari, etc.  I need to look into the
detail...

-- Yoshiki