Error code mixup?

Peter Eisentraut <peter_e@gmx.net> writes:

My copy of SQL99 assigns
most specific type mismatch 2200G
null value, no indicator parameter 22002
but elog.h has it set up the other way around. Can someone clear this up
for me?

Hoo, that's interesting. I believe that I actually built the original
version of errcodes.h by editing the list of codes in the Ada-binding
part of the spec (part 2 13.4 rule 2e), which includes, in my draft copy,

DATA_EXCEPTION_NULL_VALUE_NO_INDICATOR_PARAMETER:
constant SQLSTATE_TYPE :="2200G";
DATA_EXCEPTION_MOST_SPECIFIC_TYPE_MISMATCH:
constant SQLSTATE_TYPE :="22002";

But I see you're right that the table in section 22.1 has it the other
way around. (Digs ... looks like the contradiction is still there in
the published spec.) I wonder if there are any other inconsistencies?

Probably we should assume that the table in section 22.1 is
authoritative.

regards, tom lane