Re: problem with new autocommit config parameter and jdbc - Mailing list pgsql-hackers

From Stephan Szabo
Subject Re: problem with new autocommit config parameter and jdbc
Date
Msg-id 20020911055840.A34783-100000@megazone23.bigpanda.com
Whole thread Raw
In response to Re: problem with new autocommit config parameter and jdbc  (snpe <snpe@snpe.co.yu>)
List pgsql-hackers
On Wed, 11 Sep 2002, snpe wrote:

> On Wednesday 11 September 2002 02:38 pm, Rod Taylor wrote:
> > > > > Why rollback.This is error (typing error).Nothing happen.
> > > > > I think that we need clear set : what is start transaction ?
> > > > > I think that transaction start with change data in database
> > > > > (what don't change data this start not transaction.
> > > >
> > > > Another interesting case for a select is, what about
> > > > select func(x) from table;
> > > > Does func() have any side effects that might change data?
> > > > At what point do we decide that the statement needs a
> > > > transaction?
> > >
> > > Function in select list mustn't change any data.
> > > What if function change data in from clause ?
> >
> > Why can't the function change data?  I've done this one a number of
> > times through views to log the user pulling out information from the
> > system, and what it was at the time (time sensitive data).
> Scenario :
> Func change data in table in form clause
> I fetch 3 (after row 1 and 2) and then change row 1
> What result expect ?

Just because the behavior is sometimes undefined by the spec doesn't mean
the construct should be disallowed. Grouped character string columns also
could have implementation-dependent behavior (which never needs to be
specified), but we don't disallow that either.




pgsql-hackers by date:

Previous
From: Michael Meskes
Date:
Subject: Re: 7.3beta and ecpg
Next
From: snpe
Date:
Subject: Re: problem with new autocommit config parameter and jdbc