According to Appendix B of the JDBC spec, ResultSet.getObject should use
the mapping shown in table B-3 to determine the object class returned.
Table B-3 shows SMALLINT mapping to Integer. In postgresql-jdbc-8.0-311
a Short is returned instead, which is causing problems for our software.
It looks like a one-line change will fix this -- in the
org.postgresql.jdbc2.AbstractJdbc2ResultSet.intern alGetObject method.

Index: AbstractJdbc2ResultSet.java
================================================== =================
RCS file:
/usr/local/cvsroot/pgjdbc/pgjdbc/org/postgresql/jdbc2/AbstractJdbc2ResultSet.java,v

retrieving revision 1.74
diff -u -r1.74 AbstractJdbc2ResultSet.java
--- AbstractJdbc2ResultSet.java8 May 2005 23:50:56 -00001.74
+++ AbstractJdbc2ResultSet.java7 Jun 2005 22:53:02 -0000
@@ -115,7 +115,7 @@
case Types.BIT:
return getBoolean(columnIndex) ? Boolean.TRUE :
Boolean.FALSE;
case Types.SMALLINT:
- return new Short(getShort(columnIndex));
+ return new Integer(getShort(columnIndex));
case Types.INTEGER:
return new Integer(getInt(columnIndex));
case Types.BIGINT:

Kevin Grittner wrote:
>
> According to Appendix B of the JDBC spec, ResultSet.getObject should use
> the mapping shown in table B-3 to determine the object class returned.
> Table B-3 shows SMALLINT mapping to Integer. In
> postgresql-jdbc-8.0-311 a Short is returned instead, which is causing
> problems for our software. It looks like a one-line change will fix
> this -- in the
> org.postgresql.jdbc2.AbstractJdbc2ResultSet.intern alGetObject method.

Thanks for the report. I've applied a fix for this, plus support for
various other types in the mapping table that getObject() didn't handle
at all (BOOLEAN, TINYINT, DECIMAL, LONGVARCHAR, LONGVARBINARY, CLOB,
BLOB) to CVS HEAD and the stable 8.0 branch.

-O

---------------------------(end of broadcast)---------------------------
TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org