> > it seems that on every commit, the cursor is closed... and re-opened with
> > new
> > variables'values
> I think that currently the only way to reuse query plans would be migrating
> of your logic to backend and using SPI prepared statements.
> > btw, as many asked me, queries are VERY simple, there is only a few
> > Each query works on one table at a time. no joins for example. Only
> massive bulk
> > work with CURSORS.
> Again, can't some of it be moved to backend, either using PL/PgSQL or C (or
> pltcl, plperl, plpython ;)
We read all the " Chapter 21. Server Programming Interface" with SPI doc.
This seems _really_ interresting, make me remember of outline statments in
1) how to find some sample code? are SPI statments can be called from
2) if prepared statments and stored execution plan exist, why can't thos be used
from any client interface or simple sql?
3) You tell us we can "move to the backend" some queries: do you mean we would
have better performances with stored functions in plpgsql?
Thanks a lot Hannu, I promise to stop soon with questions :-)
This is _so_ important for us..
Best regards & wishes.
In response to
pgsql-hackers by date
|Next:||From: Rod Taylor||Date: 2002-02-28 13:56:36|
|Subject: Re: Arch (was RE: Refactoring of command.c )|
|Previous:||From: Hannu Krosing||Date: 2002-02-28 13:24:15|
|Subject: Re: Oracle vs PostgreSQL in real life|