Re: Practical sets of SQLSTATE values?

Joe Conway <mail(at)joeconway(dot)com> writes:
> Tom Lane wrote:
>> Anyone have lists of implementation-defined SQLSTATEs for
>> the big commercial DBs?

> Does this help?
> http://www.csis.gvsu.edu/GeneralInfo/Oracle/appdev.920/a97269/pc_09err.htm#3174

Some, but the mapping table is mostly pretty vague --- for instance,
it's quite unclear whether they provide subclasses in the 42xxx series
errors, or if they all come out as 42000. If there are subclasses,
which subclass codes correspond to which ORA-foo codes? A more complete
mapping table would help.

The same site has a copy of the complete Oracle 9i error message book:
http://www.csis.gvsu.edu/GeneralInfo/Oracle/server.920/a96525/toc.htm
but I couldn't find any mention at all of SQLSTATE codes in it. It's
pretty clear that Oracle regards SQLSTATE as an ugly stepchild. (Which
might be a fair assessment ;-), but it's at least somewhat standard ...)
We might do better following DB2's lead.

regards, tom lane