Here is the output of the block of code bellow.

TEST 1
u8??
[117, 56, -77, -13, 62, 75, 111, 109, 46, -52, -70, -42, 111, -35, -94, 82]

TEST 2
u8??
Exception java.sql.SQLException - Data Type Mismatch

TEST 3
[B@601bb1
[117, 56, -77, -13, 62, 75, 111, 109, 46, -52, -70, -42, 111, -35, -94, 82]

The exception when access the field from the CachedRowSet using getBytes() is "Data Type Mismatch"

Here is the code. Hope it keeps the format...

What i think it is happening is jaybird is treating it as VARCHAR, so it does some charset convertions and trunc it to 4 chars (as expected), but the implementation of CachedRowSet sees it is a VARCHAR and get the value using getString(), so it is losing part of the data. If a cast the field to binary (BLOB SUB_TYPE 0), jaybird treats the field as binary and the cachedrowset imports it right.

Class.forName("org.firebirdsql.jdbc.FBDriver");

Properties p = new Properties();
p.setProperty("charSet", "UTF-8");
p.setProperty("userName", "SYSDBA");
p.setProperty("password", "masterkey");

Connection c;
PreparedStatement s;
ResultSet r;
CachedRowSet crs;

c = DriverManager
.getConnection("jdbc:firebirdsql:192.168.0.1:teste", p);

// DIRECT ACCESS TO JAYBIRD RESULTSET
s = c
.prepareStatement("select first 1 a.id from v3$nfes_requisicoes a");
r = s.executeQuery();
r.next();
try {
System.out.println("\nTEST 1");
System.out.println(r.getString("id"));
System.out.println(Arrays.toString(r.getBytes("id")));
} catch (Exception e) {
System.out.println("Exception " + e.getMessage());
}

// ACCESS THROUGH CACHEDROWSET
s = c
.prepareStatement("select first 1 a.id from v3$nfes_requisicoes a");
r = s.executeQuery();
crs = new CachedRowSetImpl();
crs.populate(r);
crs.next();
try {
System.out.println("\nTEST 2");
System.out.println(crs.getString("id"));
System.out.println(Arrays.toString(crs.getBytes("id")));
} catch (Exception e) {
System.out.println("Exception " + e.getMessage());
}

// ACCESS THROUGH CACHEDROWSET FORCING THE FIELD TO BINARY
s = c
.prepareStatement("select first 1 cast(a.id as blob sub_type 0) as id from v3$nfes_requisicoes a");
r = s.executeQuery();
crs = new CachedRowSetImpl();
crs.populate(r);
crs.next();
try {
System.out.println("\nTEST 3");
System.out.println(crs.getString("id"));
System.out.println(Arrays.toString(crs.getBytes("id")));
} catch (Exception e) {
System.out.println("Exception " + e.getMessage());
}

> Which is correct :)

I´m not so sure about this ;-)
I think OCTETS shoud be treated as binary, as it is intended to stored unencoded bytes, like binary blobs (or am i misundertanding OCTETS purposes?)

Regards,

Fabiano

> > The problem is that i use a CachedRowSet implementation to store the ResultSet in memory and close the connection, and the CachedRowSet implementation thinks it is a VARCHAR column when caching it, and it loses information and gives an exception when i call this column from the CachedRowSet.
>
> How do you try to get the column data (rs.getString() or rs.getBytes())
> and what exception do you get?
>
> > Is this expected or can be considered a bug or a candidate for improvement?
>
> At the moment, I'd say, it is correct behaviour, but please post the
> exception you get.
>
> Roman
>