Understanding GroupBY - sql-server

Question

I have a table in SQL Server where personal entries and exits are recorded.
There are multiple hours of entry or exit on the same day. What I need is to recover the first entry and the last exit of the day.
Date hour Clock
------------------------
01/01/2017 09:00 1
01/01/2017 11:30 2
01/01/2017 17:00 2
02/01/2017 7:59 1
02/01/2017 16:00 1
I have this SQL query that works correctly.
SELECT
d.Date,
MIN(d.hour) as Entry,
MAX(dt.hour) as Exit
FROM
#temp1 AS d
LEFT JOIN
#temp1 AS dt ON d.Date = dt.Date
GROUP BY
d.Date
ORDER BY
Date DESC
BUT if I add 2 more columns to the query
SELECT
d.Date,
d.clock as ClockEntry, -- Aggregated column to display
MIN(d.hour) as Entry,
dt.clock as ClockExit, -- Aggregated column to display
MAX(dt.hour) as Exit
FROM
#temp1 AS d
LEFT JOIN
#temp1 AS dt ON d.Date = dt.Date
GROUP BY
d.Date
ORDER BY
Date DESC
I get this error:
Column '# temp1.clock' is invalid in the select list because it is not contained in either an aggregate function or the GROUP BY clause.
I just need to group by the field "date", I do not want to add more conditions to the GROUP BY.. How could I solve it?
I want this result
DATE ClockEntry Entry ClockExit Exit
-------------------------------------------------------
01/01/2017 1 09:00 2 17:00
02/01/2017 1 7:59 1 16:00

Answer 1

So there is an easy way to do this - use 2 ranking functions:
Partitioned by date, Ordered by the hour ascending
Partitioned by date, Ordered by the hour descending
At that point the rows can be joined where they both have a value of 1, and the dates match.
I tend to use a CTE for this:
with temp2 (MinId, MaxId, Date, Hour, Clock)
AS
(
select ROW_NUMBER() Over (partition by date order by hour),
ROW_NUMBER() Over (partition by date order by hour desc),
*
from temp1
)
select distinct
d1.Date,
d1.Clock,
d1.Hour,
d2.Clock,
d2.Hour
FROM temp2 d1
LEFT JOIN temp2 d2
ON d1.Date = d2.Date -- dates match
AND d1.MinId=d2.MaxId -- minId=earliest record MaxId=latest record
WHERE d1.MinId=1

Answer 2

GROUP BY Method
If you know that only a single value will get out the clock column,
just aggregate the values with a MAX or MIN aggregate function for example like this :
SELECT
d.Date,
MIN(d.clock) as ClockEntry,
MIN(d.hour) as Entry,
MAX(dt.clock) as ClockExit,
MAX(dt.hour) as Exit
FROM #temp1 AS d
LEFT JOIN #temp1 AS dt
ON d.Date= dt.Date
GROUP BY d.Date
order by Date desc
Or if you have multiple clock values and want to see them all,
add them to the GROUP BY statement :
SELECT
d.Date,
d.clock as ClockEntry,
MIN(d.hour) as Entry,
dt.clock as ClockExit,
MAX(dt.hour) as Exit
FROM #temp1 AS d
LEFT JOIN #temp1 AS dt
ON d.Date= dt.Date
GROUP BY d.Date, d.clock, dt.clock
order by Date desc
ORDER BY Method
Use a cursor or a Common Table Expression for each date to get both first entry and last exit.
First Entry for a given date
SELECT TOP 1 d.date, d.clock as ClockEntry, d.hour as Entry
FROM #temp1 AS d
WHERE d.date = #myDate
ORDER BY d.hour ASC
Last Exit for a given date
SELECT TOP 1 d.date, d.clock as ClockExit, d.hour as Exit
FROM #temp1 AS d
WHERE d.date = #myDate
ORDER BY d.hour DESC
Reference :
GROUP BY Documentation
AGGREGATE FUNCTIONS