Thread: ordering problems ...

Hello everyone,

I have two tables :

question_response and question_response_answer. 

There is a one to many relationship between the question_response and a
question_response_answer (schema below at the bottom).

When I execute :


select answer from question_response_answer qra, question_response qr 
WHERE qr.response_id = qra.response_id AND qr.personal_id =
'XXYYTEST2' AND qr.module_code = 'TEST990' AND qr.assessment_id = 109
AND qr.question_id = 125 ;

I get :
answer
---------Foo BarFredNoYes


When I execute :                                                                        
select * from question_response_answer;

I get :

response_id |   answer
-------------+------------        148 | Yes        148 | No        148 | Fred        148 | Foo Bar

I'd like to know why the ordering has changed, and is there anything I can do make it the same as it is on the "select
*from question_response_answer" "version" ?
 

Running Postgres 7.3.2 on Linux.

Schema :   Column     |         Type         |                       Modifiers
---------------+----------------------+--------------------------------------------------------response_id   | integer
           | not null default
 
nextval('question_reponse_seq'::text)question_id   | integer              |personal_id   | character varying(9)
|assessment_id| integer              |module_code   | character varying(9) |
 
Indexes: question_response_pkey primary key btree (response_id),        question_response_question_id_key unique btree
(question_id,
personal_id, assessment_id)

(there are foreign keys, but I've left them off)

tweektest=> \d question_response_answer;
Table "public.question_response_answer"  Column    |  Type   | Modifiers
-------------+---------+-----------response_id | integer | not nullanswer      | text    | not null
Indexes: question_response_answer_pkey primary key btree (response_id,
answer)
Foreign Key constraints: $1 FOREIGN KEY (response_id) REFERENCES
question_response(response_id) ON UPDATE NO ACTION ON DELETE CASCADE




Thanks in advance,
David.

-- 
David Goodwin 

[ dof at codepoets dot co dot uk   ]
[ http://www.codepoets.co.uk       ]

DIB Walid wrote :
> Hello,
> I think that Postgres use the default ORDER BY function:
> so in your first request there is a default order by on the "answer" 
> In order to obtain the same order in your second request, put"ORDER BY 
> ANSWER" in your request, because if you miss this point, postgres will order 
> by id automatically, i hope i understand your problem because my english is a 
> little weak.
> @+
> Walid
> INRIA Rhones Alpes

Hi,

Thanks for the speedy reply :)

I just discovered that if i put in a "order by qra.oid" it works ok.
If I run the query with "select * from ..." rather than "select answer
from .." the results are the same (ordered alphabetically by answer).

thanks
David.
-- 
David Goodwin 

[ dof at codepoets dot co dot uk   ]
[ http://www.codepoets.co.uk       ]

David Goodwin <dof@codepoets.co.uk> writes:
> I just discovered that if i put in a "order by qra.oid" it works ok.

That isn't going to work for long --- for example, as soon as you update
any row in question_response_answer, physical order will stop matching
OID order.

The real answer to your question is that SQL does not consider the
ordering of a table to be significant.  If you want the rows of a query
result to be displayed in a particular order, you have to use ORDER BY.
Without that, the system is entitled to deliver the rows in any order
it finds convenient.
        regards, tom lane