> It seems that slowly, slowly the CG guys begun to write about the
> challenges that may appear due the Unicode support which will
> be present
> in Tiburon. There's a thread now in CodeGear's .non-technical
> newsgroup
> where Allen Bauer (CG chief scientist) has a post and, also, one of
> their R&D engineers (btw, nice guy) posted some things to take care at

http://feeds.feedburner.com/~r/ChrisBensenBlog/~3/185430522/unicode-sizeof-i
s-different-than-length_15.html

> Do you are aware of them?

I was not aware of these specific details but certainly the idea of having
physical size vs. character size interchangability goes away with multi-byte
character sets. Seems that this has already bitten the Firebird team where
UTF-8 handling is concerned. I am still not yet aware of how physical byte
overruns are going to be handled when by logical definition a 20 character
varchar string requires 21 physical bytes to be represented but the client
buffer only has up to 20 bytes allocated for its storage. What I foresee,
at least in the short-term is either corrupted strings or exceptions when it
is being transliterated when someone is entering what they believe to be a
string well within the constraints of the column.

Thanks for drawing my attention to this conversation.

Kind regards,
Jason Wharton