Oliver,
Martin's existing implementation handles a common set of usage for Symbol
fonts, where charsets are ignored, and no attempt made to actually do a
direct mapping to Unicode.
It's just a presentation mode where, say, the glyph for the codepoint which
corresponds to capital B may be rendered as a greek B. This is all that,
say, StarOffice does, and is a subset of what Word does. When using
non-Unicode fonts, this is a defensible approach.
To get a more Unicode-savvy implementation of the Symbol dialog would
require additional platform-specific code to detect which characters are
actually encoded in each font, preferably segregated by subsets. (For two
examples of this approach, see the NT character map, or more recent versions
of Word.)
This isn't necessarily rocket science, but it's not an overnight task
either.
Paul
