If I recall, it means that for a given query select ID from table1 t1,
table2 t2 where t1.id = t2.id, you will have to do something like
this:
List list = select T1_ID from table1;
for each item in list
{
select * from Table2 where t2.id = item.t1_id
}
The multiplication is number of the queries issued, not the memory used.
Obviously, the simple case above could be optimized using an IN
statement, but anything that wasn't trivial could not be.
On 6/20/07, Øyvind Harboe <[EMAIL PROTECTED]> wrote:
In the link below it says that relationships does not work across
databases "as it will require Cayenne to do in memory cartesian
product processing".
What does this qualifier mean? Does it mean that it is possible to
enable this, but it would take a lot of memory? That this feature was
considered, but rejected as it would have not worked in all cases?
For the cases that we need this, there is no problem for Cayenne to
have both tables completely in memory (always < 1000 records) so I'm
not concerned about CPU processing power or memory consumption.
Has any work been done on supporting this?
We've got a real database (Derby) that has relationships to other
existing datbases(MS Access/FoxPro databases). We can't move the
existing databases into Derby due to bug-by-bug compatibility
problems.
http://cwiki.apache.org/CAY/multiple-databases-example.html
--
Øyvind Harboe
http://www.zylin.com - eCos ARM & FPGA developer kit
