Re: Queries with large ResultSets

From: Andrea Aime <andrea(dot)aime(at)aliceposta(dot)it>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: Kris Jurka <books(at)ejurka(dot)com>, Jack Orenstein <jao(at)geophile(dot)com>, pgsql-jdbc(at)postgresql(dot)org
Subject: Re: Queries with large ResultSets
Date: 2004-05-20 19:27:11
Message-ID: 200405202127.11956.andrea.aime@aliceposta.it
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-jdbc

Alle 20:44, giovedì 20 maggio 2004, Tom Lane ha scritto:
> Andrea Aime <andrea(dot)aime(at)aliceposta(dot)it> writes:
> > Ugh... those limitation are really frightening, this means we cannot
> > fetch big quantities of data outside of a transaction...
>
> You were expecting something different? Postgres does not do *anything*
> outside of a transaction.

Every other database I have some experience on (sql server, sapdb) allows you
to use cursors regardless of the transaction. If you are working with a
database in autocommit mode that doesn't mean that you don't need to load
huge quantities of data... on the contrary, I would expect that it would be
more costly to load the huge amount of data inside of a transaction because
of transaction isolation.
Anyway, that's not the matter, the real problem is that our generic jdbc code
won't work properly with postgres because of this "feature" thus we will have
to subclass everything that deals with the database in order to get the
correct behaviour.

If I am in autocommit = false mode, I will have to call connection.commit()
after every write command to the database... isolation wise, to simulate a
true autocommit, will I have to call connection.commit() also after reads if
I raise the isolation level above READ_COMMITED?

Best regards
Andrea Aime

In response to

Responses

Browse pgsql-jdbc by date

  From Date Subject
Next Message Dave Cramer 2004-05-20 20:03:33 Re: Queries with large ResultSets
Previous Message Tom Lane 2004-05-20 18:44:24 Re: Queries with large ResultSets