aggregate functions are not allowed in WHERE - sql

Question

I am using this query to find the unique records by latest date using postgresql. The error I am having is "aggregate functions are not allowed in WHERE". How to fix error “aggregate functions are not allowed in WHERE” Following this link I have tried to use inner select function. But this did not work. Please help me to edit the query. I am using PgAdmin III as client.
SELECT Distinct t1.pa_serial_
,t1.homeownerm_name
,t1.districtvdc
,t1.date as firstrancheinspection_date
,t1.status
,t1.name_of_data_collector
,t1.fulcrum_id
,first_tranche_inspection_v2_reporting_questionnaire.date_reporting
From first_tranche_inspection_v2 t1
LEFT JOIN first_tranche_inspection_v2_reporting_questionnaire ON (t1.fulcrum_id = first_tranche_inspection_v2_reporting_questionnaire.fulcrum_parent_id)
where first_tranche_inspection_v2_reporting_questionnaire.date_reporting = (
select Max(first_tranche_inspection_v2_reporting_questionnaire.date_reporting)
from first_tranche_inspection_v2
where first_tranche_inspection_v2.pa_serial_ = t1.pa_serial_
);

Answer 1

You want to join the latest reporting questionaire per inspection. In PostgreSQL you can use DISTINCT ON for this:
select fti.*, rq.*
from first_tranche_inspection_v2 fti
left join
(
select distinct on (fulcrum_parent_id) *
from first_tranche_inspection_v2_reporting_questionnaire
order by fulcrum_parent_id, date_reporting desc
) rq on rq.fulcrum_parent_id = fti.fulcrum_id;
Or use standard SQL's ROW_NUMBER:
select fti.*, rq.*
from first_tranche_inspection_v2 fti
left join
(
select
ftirq.*,
row_number() over (partition by fulcrum_parent_id order by date_reporting desc) as rn
from first_tranche_inspection_v2_reporting_questionnaire ftirq
) rq on rq.fulcrum_parent_id = fti.fulcrum_id and rq.rn = 1;
What you were trying to do should look like this:
select fti.*, rq.*
from first_tranche_inspection_v2 fti
left join first_tranche_inspection_v2_reporting_questionnaire rq
on rq.fulcrum_parent_id = fti.fulcrum_id
and (rq.fulcrum_parent_id, rq.date_reporting) in
(
select fulcrum_parent_id, max(date_reporting)
from first_tranche_inspection_v2_reporting_questionnaire
group by fulcrum_parent_id
);
This works, too, and only has the disadvantage that you read the table first_tranche_inspection_v2_reporting_questionnaire twice.

Answer 2

DISTINCT often ends up being implemented with a GROUP BY query in many RDBMS. What I think is happening in your current query is that there is already an implicit aggregation involving the columns in your SELECT. Hence, the correlated subquery involving MAX() actually is an aggregation because of the DISTINCT.
One quick workaround might be to perform the original query without DISTINCT, then subquery the result set to retain only distinct records:
WITH cte AS (
SELECT t1.pa_serial_,
t1.homeownerm_name,
t1.districtvdc,
t1.date as firstrancheinspection_date,
t1.status,
t1.name_of_data_collector,
t1.fulcrum_id,
t2.date_reporting
FROM first_tranche_inspection_v2 t1
LEFT JOIN first_tranche_inspection_v2_reporting_questionnaire t2
ON t1.fulcrum_id = t2.fulcrum_parent_id
WHERE t2.date_reporting = (SELECT MAX(t.date_reporting)
FROM first_tranche_inspection_v2 t
WHERE t.pa_serial_ = t1.pa_serial_)
);
SELECT DISTINCT t.pa_serial_,
t.homeownerm_name,
t.districtvdc,
t.firstrancheinspection_date,
t.status,
t.name_of_data_collector,
t.fulcrum_id,
t.date_reporting
FROM cte t
Note that I went ahead and added an alias to the second table in your join, which leaves the query much easier to read.