Re: [SQL] Odd "problem", not sure if there is a solution ....

secret wrote:

> The Hermit Hacker wrote:
>
> > am working up a database for client that contains answers for online
> > testing...right now, the 'tables' look like:
> >
> > userid Q1 Q2 Q3 Q4 Q5 Q6 Q7 etc
> >
> > I want to change it so that its more generic, as:
> >
> > userid question_no answer
> >
> > so that instead of one row per user, there are X rows...
> >
> > the problem is that, somehow, I want the client to be able to view the
> > results as the first format (ie. one line per user) with it being stored
> > as multiple lines per user...
> >
> > Possible?
> >
> > Marc G. Fournier ICQ#7615664 IRC Nick: Scrappy
> > Systems Administrator @ hub.org
> > primary: scrappy(at)hub(dot)org secondary: scrappy(at){freebsd|postgresql}.org
>
> Sure it's possible, if you can live with the constrant that every person must
> answer every question... (Else that person will kind of 'disappear' <cough>):
>
> CREATE TABLE q (
> person int4, -- Person ID#
> q int4, -- Question #
> a int4 ) -- Answer
>
> ftc=> select * from q where q in (1,2);
> person|q| a
> ------+-+--
> 1|1| 1
> 1|2| 1
> 2|1|21
> 2|2|22
> (4 rows)
> SELECT q1.person,
> q1.a,
> q2.a
>
> FROM q as q1,
> q as q2
>
> WHERE
> q1.person=q2.person
> AND q1.q = 1
> AND q2.q = 2
> ;
> person| a| a
> ------+--+--
> 1| 1| 1
> 2|21|22
> (2 rows)
>
> Your probably want to rename "a" to a1 and a2, but you can expand this to N
> questions... The performance will suck if your table is pretty big, so I'd arrange
> to move things to a history file every so often... Keep in mind people will
> disappear if they are missing the answer to any question.

Hi David,

I like your solution.
However, is there a better way to remove the constraint?

Thanks

Regards,
Chai