On 17/02/2010, at 1:00 AM, Joseph Wright wrote:

> My understanding was that the "native" encoding, at least for XeTeX, was UTF-8.

I'm not aware of any encoding difficulties in XeTeX.
For example, one can write

    \let\a=𝔄
    [\a]

and receive the log message

    Missing character: There is no character number "1D504 in font ...

IIRC, at one stage XeTeX was restricted to UTF-16 input chars (or at least that \char and friends were restricted to 16 bits), but that changed when JK realised that we needed full UTF-32 support for maths.

-- Will