From: | Richard Poole <richard(dot)poole(at)vi(dot)net> |
---|---|
To: | pgsql-sql(at)postgresql(dot)org |
Subject: | Re: count(*) - join question |
Date: | 2001-11-27 14:25:30 |
Message-ID: | 20011127142530.C24565@office.vi.net |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-sql |
On Mon, Nov 26, 2001 at 11:03:56PM +0000, Nicole S. Weber wrote:
>
> I have tried to use a join on the sample_id in combination with
> COUNT(*)/GROUP BY sample_id:
>
> SELECT animals.sample_id, samples.sample_date, samples.sample_code,
> count(*)
> FROM animals, samples
> WHERE samples.sample_id = animals.sample_id
> GROUP BY animals.sample_id;
>
> Alas, this returns:
>
> ERROR: Attribute samples.sample_Dadate must be GROUPed or used in an
> aggregate function
You need to GROUP BY all the columns that you're not counting the
different occurrences of:
SELECT animals.sample_id, samples.sample_date, samples.sample_code, count(*)
FROM animals, samples
WHERE samples.sample_id = animals.sample_id
GROUP BY animals.sample_id, samples.sample_date, samples.sample_code;
Richard
From | Date | Subject | |
---|---|---|---|
Next Message | Aasmund Midttun Godal | 2001-11-27 19:53:51 | Re: PL/pgSQL examples NOT involving functions |
Previous Message | Gyorgy Molnar | 2001-11-27 14:11:18 | Queue in SQL |