On the other hand, maybe those drivers are following the spec. From
PreparedStatement javadocs:
"void setObject(int parameterIndex, Object x, int targetSqlType)
throws SQLException
Sets the value of the designated parameter with the given object. This
method is like the method setObject above, except that it assumes a
scale of zero."
So truncating a value to scale = 0 is actually appropriate if the
scale is not mapped. Of course something like 'setBigDecimal' would
probably preserve it, so maybe we still need a special ExtendedType to
call this method.
Andrus
On Aug 8, 2008, at 2:28 PM, Andrus Adamchik wrote:
> Here is a potential hole. By default (unless ExtendedType is
> explicitly defined for a given Java class), Cayenne does the
> following JDBC calls:
>
> PreparedStatement st = ..
> if (scale != -1) {
> st.setObject(pos, val, type, scale);
> }
> else {
> st.setObject(pos, val, type);
> }
>
> Looks correct from the JDBC standpoint, but from my experience some
> drivers may have it all wrong, and inconsistent with a corresponding
> 'setSpecificType(..)' method for a given object class. So what is
> the class of the object parameter? BigDecimal? Something else?
>
> Andrus
>
>
> On Aug 8, 2008, at 2:16 PM, Øyvind Harboe wrote:
>
>> On Fri, Aug 8, 2008 at 7:53 PM, Andrus Adamchik <andru..bjectstyle.org
>> > wrote:
>>> Cayenne doesn't do anything special on Derby vs. SQLServer in this
>>> respect.
>>
>> My thinking was that databases behaved differently here and that
>> Cayenne then either defines that the default behaviour is undefined
>> when scale is not specified or that Cayenne defines it to be the
>> same as whatever the database does.
>>
>>> Is this behavior happening for selected objects or when binding a
>>> query
>>> parameter?
>>
>> First I run an update w/a prepared statement(or rather Cayenne uses
>> prepared
>> statements for updates) and afterwards the queries return e.g. "1"
>> instead of "1.5".
>>
>> If I run a select statement from *outside* Cayenne, it reveals that
>> the data in
>> the database are actually truncated.
>>
>> Is there a testcase for this in Cayenne?
>>
>> --
>> Øyvind Harboe
>> http://www.zylin.com/zy1000.html
>> ARM7 ARM9 XScale Cortex
>> JTAG debugger and flash programmer
>>
>
>
This archive was generated by hypermail 2.0.0 : Fri Aug 08 2008 - 14:35:44 EDT