On Thu, 15 Nov 2007, Lance Andersen wrote:
>
> The following bug has been logged online:
>
> Bug reference: 3751
> PostgreSQL version: 8.2.x
> Description: Conversion error using PreparedStatement.setObject()
> Details:
>
> A PreparedStatement.setObject(1, "1", Types.Bit) will fail with the
> following Exception:
This is not a great test case.
1) It doesn't contain the definition of the Drivers class so it doesn't
compile. The whole drivers class and dumping metadata is needless
complication for a simple test case.
2) It doesn't contain the definition of the bit_tab table, so it doesn't
run.
3) The error is actually coming from "stmt.executeUpdate(Min_Insert)", not
a PreparedStatement.
So where does that leave us?
1) The raw insert fails. INSERT INTO bit_tab (1,0,null) fails because 1
gets typed as an integer and there are no implicit or assignment casts
from integer to bit. You would need to say, '1' so it comes in untyped
and converted to bit, or be explicit about the type with a cast, saying
1::bit or CAST(1 AS bit).
2) There might be a problem with bit conversion in prepared statements,
but we didn't get that far.
Other notes:
In PG the bit type is really for multiple bits, not a single
bit. Consider SELECT 77::bit(8) results in "01001101". It's more
likely that you want to use boolean as the type instead although it
doesn't have any casts that will help you out in this situation either.
Kris Jurka
