Re: PQftable insufficient for primary key determination - Mailing list pgsql-general

From mmc@maruska.dyndns.org (Michal Maruška)
Subject Re: PQftable insufficient for primary key determination
Date
Msg-id m27jt4fx9k.fsf@linux11.maruska.tin.it
Whole thread Raw
In response to Re: PQftable insufficient for primary key determination  (Bruno Wolff III <bruno@wolff.to>)
Responses Re: PQftable insufficient for primary key determination  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-general
Bruno Wolff III <bruno@wolff.to> writes:

> On Fri, Jul 16, 2004 at 03:38:52 +0200,
>   Michal Maru?ka <mmc@maruska.dyndns.org> wrote:
>>
>>
>> Imagine i have a table A with primary key "p" and another attribute called "data":
>>
>> SELECT * from A, A, B;
>>
>> here, i will have 2 columns "data", PQftable tells me that they are from A, but
>> which of the 2 columns "p" is the primary key of the tuple (of the table A)?
>>
>>
>> Is it possible to get distinguish the 2 instances of the table A?



> You can use aliases to distinguish between two references to the same
> table in a query. Your question doesn't make a lot of sense though, since
> you haven't said which of the two copies of A you are interested in.

i don't want to force the user to distinguish 'manually', nor depend on it.

EXPLAIN VERBOSE {query}  seems to provide that information, if i walk down
:varno, i *guess*. I haven't found any reference documentation  on the output of
the EXPLAIN VERBOSE.


So, my question is: should i look at the code which walks that tree
(probably related to the function SendRowDescriptionMessage), or
is this code (which provides the distinguishing info) already available?



pgsql-general by date:

Previous
From: Bruno Wolff III
Date:
Subject: Re: PQftable insufficient for primary key determination
Next
From: Tom Lane
Date:
Subject: Re: PQftable insufficient for primary key determination