On Mon, Sep 20, 2010 at 5:27 PM, Merlin Moncure <mmoncure@gmail.com> wrote:
On Fri, Sep 17, 2010 at 3:59 PM, Sam Mason <sam@samason.me.uk> wrote: > On Fri, Sep 17, 2010 at 04:51:46PM +0200, Willy-Bas Loos wrote: >> i have a function that produces a result in xml. >> that is one row, one value even, but it grows pretty large. >> how is that handled? > > Rows are sent back in the entireity, so the PG instance would need > enough memory to work with that row. When you're running a 32bit > version of PG, values whose size is beyond ~100MB are a bit touch and go > whether it will work.
Is'nt "work_mem" used for this on the server side (when sorts and hash-tables are required) and "temp_buffers" (when temporary tables are created) ?