complex query to pull unique values

[John W. Vinson]

I have an inspection table to keep track of rooms in various buildings.
These rooms are shared with different supervisors. One inspection of a
location would result in x # of inspection records. If the room was shared
by 3 supervisors, the same inspection would generate 3 inspection records.
Comment field on each record would be different according to the supervisor,
were their employees following protocol, safety issues etc.

Since there are so many fields, trying to pull unique values is difficult,
so I made a new field in the query which concatenates SupervisorID, Bldg &
Room to make a single unique value. This part is fine.

Fine... but unnecessary. You can group by up to TEN fields; it is neither
necessary nor particularly helpful to create a redundant concatenated field.

I can pull unique Supervisor/location records with no problem. However, I
only want the most recent inspection. I set the Totals part of the
concatenated field to "Group By" and the rest to "First". Even though the
dates are sorted as descending in the query, it pulls the First date from
the original table, which by default is sorted ascending. So essentially I
get the most outdated inspection, not the most recent.

First is misleading. It returns the first record *in disk storage order* -
essentially an arbitrary, meaningless record. A Subquery with a criterion
such as

=(SELECT Max([datefield]) FROM tablename AS X WHERE X.SupervisorID =
tablename.SupervisorID)

will be a better approach.

I've tried a two query approach. Make the first query, just sort the dates
descending. Then make a new query based on the first, with the concatenated
fields and so on. It gives the same result.

Because of the same problem - FIRST isn't the "first" in the way you would
think.

 

[John W. Vinson]

Same query, except the for the date instead of First, I put Last and it
works.

sheer coincidence and good luck. Use the subquery instead.

 

[John W. Vinson]

in the query view or sql view?

You can use either, but in the query grid (which is NOT the query, just a tool
to help build SQL) you'll need to put the subquery - as a SQL string in
parentheses - in the Criteria box. You'll need to adapt the fieldnames and
tablenames to match your actual tables, which I cannot see.

 

[JR]

I have an inspection table to keep track of rooms in various buildings.
These rooms are shared with different supervisors. One inspection of a
location would result in x # of inspection records. If the room was shared
by 3 supervisors, the same inspection would generate 3 inspection records.
Comment field on each record would be different according to the supervisor,
were their employees following protocol, safety issues etc.

Since there are so many fields, trying to pull unique values is difficult,
so I made a new field in the query which concatenates SupervisorID, Bldg &
Room to make a single unique value. This part is fine.

I can pull unique Supervisor/location records with no problem. However, I
only want the most recent inspection. I set the Totals part of the
concatenated field to "Group By" and the rest to "First". Even though the
dates are sorted as descending in the query, it pulls the First date from
the original table, which by default is sorted ascending. So essentially I
get the most outdated inspection, not the most recent.

I've tried a two query approach. Make the first query, just sort the dates
descending. Then make a new query based on the first, with the concatenated
fields and so on. It gives the same result.

I've tried putting my date field as the first column in the query. It gives
the same result.

How can I solve this date sorting issue?

 

[JR]

Of course 5 minutes after I post, I found a solution but I'm not sure how it
works.

Same query, except the for the date instead of First, I put Last and it
works. I get the most recent inspection record per Supervisor/Room combo.
Great.

But when I look at my query results, Inspection ID = 2, but its date is not
the same date Inspection ID = 2 is in the table.

Inspection ID is pk for the table, autonumbered.

Query result: InspectionID = 2, Name/Room = Smith 123, Date = January 6,
2010
Table row: InspectionID = 2, Name/Room = Smith 123, Date = June 6, 2009


So if i want to go back to see the details from that particular inspection
(the most recent of Supervisor/Room), the Inspection ID is incorrect.

Not quite gettin' it........

 

[JR]

in the query view or sql view?

sheer coincidence and good luck. Use the subquery instead.

 

[JR]

Absolute brilliance!

Thanks so much, John.

You can use either, but in the query grid (which is NOT the query, just a
tool
to help build SQL) you'll need to put the subquery - as a SQL string in
parentheses - in the Criteria box. You'll need to adapt the fieldnames and
tablenames to match your actual tables, which I cannot see.