Re: Out of memory error on huge resultset

Dave Cramer <dave@micro-automation.net>

From: Dave Cramer <Dave@micro-automation.net>
To: Michael Paesold <mpaesold@gmx.at>
Cc: Aaron Mulder <ammulder@alumni.princeton.edu>, pgsql-jdbc <pgsql-jdbc@postgresql.org>
Date: 2002-10-11T14:48:41Z
Lists: pgsql-hackers, pgsql-general
Michael,

You are correct, commit will effectively close the cursor. 

This is the only way to deal with large result sets however.

Dave
On Fri, 2002-10-11 at 10:44, Michael Paesold wrote:
> Dave Cramer <Dave@micro-automation.net> wrote:
> 
> > No disadvantage, in fact that is what we would like to do.
> >
> >
> > setFetchSize(size)  turns on cursor support, otherwise fetch normally
> >
> > Dave
> >
> > On Fri, 2002-10-11 at 10:30, Aaron Mulder wrote:
> > > What would be the disadvantage of making the JDBC driver use a
> > > cursor under the covers (always)?  Is it significantly slower or more
> > > resource-intensive than fetching all the data at once?  Certainly it
> seems
> > > like it would save memory in some cases.
> > >
> > > Aaron
> 
> Well, using a cursor based result set *always* is not going to work. Cursors
> will not be held over a commit, whereas a buffer result set will. So the
> setFetchSize..
> 
> Regards,
> Michael Paesold
> 
>