On Wed, 11 Oct 2000, Joseph Shraibman wrote:
> Peter Mount wrote:
> >
> > On Wed, 11 Oct 2000, Michael Stephenson wrote:
> >
> > > Two things.
> > >
> > > Firstly, when dealing with a large ResultSet (about 120000 rows), I get a
> > > null pointer exception on the line:
> > > wasNullFlag = (this_row[columnIndex - 1] == null);
> > > Whenever I call getString(), has anyone else had this? And does anybody
> > > have a solution?
> >
> > Are you getting any out of memory errors at all?
> >
> > The problem with the current implementation is that it reads the entire
> > result into memory, so 120000 rows may be filling up your VM's memory
> > (defaults to about 16Mb).
> >
> > Does it occur if you add the -mx argument to java, ie:
> >
> > java -mx 64m uk.org.retep.sql.RetepSQL
> >
> > I'm in the design stage of implementing a version of ResultSet that will
> > use cursors, to limit how much is loaded in memory at a time.
> >
>
> The problem with that is the same problem I ran into with locking.
> cursors need to be in a BEGIN END block, and other calls from different
> Statement objects using the same db connectioni will interfere. Make
> sure it is clearly documented that that will not be thread safe (unless
> postgres gets named locks by then).
Yes, there is a problem with multiple statements and transactions on the
same connection when it comes to thread safety.
The problem as I see it is two fold.
1: How to deal with two statements within one transaction. Related to this
is the metadata methods that issue queries, how to deal with them while
within a transaction.
2: Currently the JDBC Specs only have transactions supported at the
Connection level, so I can't see how they thought that Statements could
possibly run within their own transactions, unless they thought that a
workaround of this is the use of batches.
Peter
--
Peter T Mount peter@retep.org.uk http://www.retep.org.uk
PostgreSQL JDBC Driver http://www.retep.org.uk/postgres/
Java PDF Generator http://www.retep.org.uk/pdf/