skoczian wrote:

>
> There is one thing more: even if the database character set uses one
> byte per character (ISO8859_something, ASCII, WINsomething...), but
> the connection character set is UTF8, the client gets 4*N bytes for a
> CHAR(N) field, padded with spaces.

Correct.

> And that doesn't only look ugly in
> a tabular representation, it can trip you up working with the data.

CHAR is not really user-friendly, I'd say. Especially with variable-byte
character sets.

> On the other hand, if the system character set is UTF8, another
> connection character set causes problems with the non-ASCII characters.

Only if it retrieves characters outside the connection character set
range. This is quite expected, don't you think so?


Dmitry