On 05/12/2012 01:11, Raymond O'Donnell wrote:
> On 05/12/2012 01:04, Kenneth Tilton wrote:
>> I am porting from MySQL some code that has to take an arbitrary query
>> involving joins and build up a dictionary (in an HLL talking to Postgres
>> over a socket) where each column name will be the key. The catch is that
>> there will be duplicate entries where two joined tables have the same
>> column such as "id", so I have to get the source table for each column.
>> Here is a sample query:
>>
>> select * from providers p inner join provider_types pt on pt.id
>> <http://pt.id> = p.provider_type_id;
>>
>> I actually figured out how to get the table OID which would suffice, but
>> I am porting MySQL code that could get fully qualified column names
>> including a table alias if that were used. We allow other code to look
>> up values in the dictionary with the alias as a prefix as a convenience,
>> eg. "p.id <http://p.id>" or "pt.id <http://pt.id>".
>
> Maybe I'm misunderstanding, but why don't you just give the columns an
> alias directly? -
>
> select p.id as p_id, pt.id as pt_id, ....
Sorry, I *am* misunderstanding - I missed the "arbitrary" bit. Too late
at night to be working...
Ray.
--
Raymond O'Donnell :: Galway :: Ireland
rod@iol.ie