>>I though, it interprets CHAR(30) like a field that can hold up to 30
>>bytes, which would -- in worst case -- lead to 15 unicode chars that
>>the field can hold. In our case it is very unlikely to have unicode
>>chars in the field and if it does, only one or two...
>
>No, char(30) in unicode would be 90 bytes but it ought to be only 30
>characters (that's if UNICODE_FSS supports 3-byte characters, as I
>think it does). There are two rdb$ columns, one for the length in
>characters, the other for the length in bytes. Check out the bug
>list, for I'm sure Claudio fixed it so that it checks character
>count now in a place where the InterBase code was erroneously
>checking the byte count.

The only thing I found in the bug list was bug 222563. It relates to isql
and others not reading the correct metadata, but shouldn't the server
itselve check wehter an insert into its tables conforms to the declaration
of the table?

Herbert