skjoldtc (4/29/2010)

---

65533 is a valid unicode number but represents a special replacement character. It is the highest value in the character set. dbowlin makes a good point about what values to check for. You would ahve to know the source and the target databases to know what's reasonable. I doubt that 65533 is reasonable under 99.9% of the cases but there may be rare instances.

Good question, though.

skjoldtc (4/29/2010)

---

65533 is a valid unicode number but represents a special replacement character. It is the highest value in the character set. dbowlin makes a good point about what values to check for. You would ahve to know the source and the target databases to know what's reasonable. I doubt that 65533 is reasonable under 99.9% of the cases but there may be rare instances.

Good question, though.

select nchar(65533); -- returns value
select nchar(65535); -- returns value
select nchar(65536); -- returns null because 65536 is obviously not valid
-- as it cannot really fit into 2 bytes

Oleg Netchaev (4/29/2010)

---

Do you know why 65533 is the highest value? Theoretically, the highest should be 65535. This is consistent with nchar implementation, i.e.

select nchar(65533); -- returns value
select nchar(65535); -- returns value
select nchar(65536); -- returns null because 65536 is obviously not valid
-- as it cannot really fit into 2 bytes

Good question, I learned something new today.

Oleg

skjoldtc (4/29/2010)

---

No. I don't know why that is.

I now recall that the AS400 has a max CCSID (coded character set ID) of 65533. That makes sense since the original data came from a mainframe. IBM mainframes and AS400 both use EBCDIC. There are some unprintable and undisplayable characters and 65533 is used as a replacement on those systems. Basically, IIRC, it ends being a printable/displayble character defined system-wide that is substituted. So, you could define that a ~ (or any other printable/displayable character) prints or displays instead of throwing an error.

At my age, that's a lot to recall, so, I may be mistaken.