Yes, you're correct. I guess this makes sense but it does seem strange
that I can enter garbage in a query but it still runs. And in my case
the output from this (the entire table) was then used in a delete
statement that toasted the entire table. Allowing bogus SQL just seems
"wrong" but I do understand what's going on.
Thanks for your help.
On 02/22/2011 10:45 AM, Tom Lane wrote:
> "Scott Dunbar"<scott(at)xigole(dot)com> writes:
>> I have a nested in clause like:
>> select respondent_id from respondent where respondent_id in (select
>> respondent_id from chat_session where project_id in (select project_id from
>> project where company_id = 4));
>> However, in this example, there is no column named respondent_id in the
>> chat_session table.
> Probably there is one in respondent, though? This behavior is not a bug
> --- what you have there is an outer reference, and it is working exactly
> as specified by the SQL standard. Sub-selects would be a whole lot less
> useful if they couldn't refer to variables of the outer query.
> regards, tom lane
Xigole Systems, Inc.
Enterprise software consulting, development, and hosting
In response to
pgsql-bugs by date
|Next:||From: Daniel Farina||Date: 2011-02-22 18:32:53|
|Subject: Re: Re: 8.3.5: Types with typnamespace pointing at
non-existent pg_namespace oid|
|Previous:||From: Tom Lane||Date: 2011-02-22 17:45:03|
|Subject: Re: BUG #5898: Nested "in" clauses hide bad column names |