That is exactly what we need for postgres. In particular JDBC could use
the ability to have a new connection to a backed that doesn't require a
new backed to use cursors for scrollable resultsets so big selects don't
have to download everything before results can be used. There are many
other reasons for having multiple connections to one backed process,
mainly to do with different transactions not screwing each other up.
On Sat, 28 Apr 2001, The Hermit Hacker wrote:
> On Fri, 27 Apr 2001, Clayton Vernon wrote:
>
> > I was wondering what the plans were for PostgreSQL to convert to the
> > one process multithreaded approach, as Apache, Interbase and others
> > are doing?
>
> there has been talk about doing some threads actions inside of a
> process, but, if I recall my read of Apache2, they are still doing
> multi-process, with threading inside of each process ...
>
>