On Sun, 1 Jul 2001, Tom Lane wrote:
> Alex Pilosov <alex(at)pilosoft(dot)com> writes:
> > I think I just understood what you were saying: having tupleDesc in RTE is
> > not kosher, because RTE can last longer than a given tupleDesc?
> Depends where you got the tupleDesc from --- if you copy it into the
> parse context then it's OK in terms of not disappearing. However,
> that doesn't mean it's still *valid*. Consider
> declare foo cursor for select * from bar;
> create view v1 as select * from cursor foo;
> Now the cursor foo is no more, but v1 still exists ... what happens
> when we try to select from v1?
I believe it will fail in the executor trying to open the portal...That's
the expected behaviour, right?
In response to
pgsql-hackers by date
|Next:||From: Vince Vielhaber||Date: 2001-07-01 23:27:25|
|Subject: Re: Now it's my turn...|
|Previous:||From: Tom Lane||Date: 2001-07-01 04:47:41|
|Subject: Re: RangeTblEntry modifications |