Datetime data type in a CASE expression impacts all other values on SQL Server 2016 - sql

Question

I'm attempting to insert a few rows into a table in SQL Server 2016 using the SQL query below. A temp table is used to store the data being inserted, then a loop is used to insert multiple rows into the table.
--Declare temporary table and insert data into it
DECLARE #fruitTransactionData TABLE
(
Category VARCHAR (30),
Species VARCHAR (30),
ArrivalDate DATETIME
)
INSERT INTO #fruitTransactionData ([Category], [Species], [ArrivalDate])
VALUES ('Fruit', 'Apple - Fuji', '2017-06-30')
--Go into loop for each FieldName (there will be 3 rows inserted)
DECLARE #IDColumn INT
SELECT #IDColumn = MIN(ID) FROM FieldNames
WHILE #IDColumn IS NOT NULL
BEGIN
--Insert data into Transactions
INSERT INTO [dbo].[Transactions] ([FieldName], [Result])
SELECT
(SELECT Name FROM FieldNames WHERE ID = #IDColumn),
CASE
WHEN #IDColumn = 1 THEN 1 --Result insert for FieldName 'Category' where ID=1 refers to 'Fruits'
WHEN #IDColumn = 2 THEN 99 --Result insert for FieldName 'Species' where ID=99 refers to 'Apple - Fuji'
WHEN #IDColumn = 3 THEN [data].[ArrivalDate] --Result insert for FieldName 'Date'
ELSE NULL
END
FROM
#fruitTransactionData [data]
--Once a row has been inserted for one FieldName, then move to the next one
SELECT #IDColumn = MIN(ID)
FROM FieldNames
WHERE ID > #IDColumn
END
When inserting the data, the data is inserted, but all the results show dates, when some data weren't meant to be dates.
+-----+------------+---------------------+
| ID | FieldName | Result |
+-----+------------+---------------------+
| 106 | Category | Jan 2 1900 12:00AM |
| 107 | Species | Apr 10 1900 12:00AM |
| 108 | Date | Jun 30 2017 12:00AM |
+-----+------------+---------------------+
If I comment out the row insert of the date, the columns display correctly.
+-----+------------+--------+
| ID | FieldName | Result |
+-----+------------+--------+
| 109 | Category | 1 |
| 110 | Species | 99 |
+-----+------------+--------+
It seems like the insertion of the date converts all the result values to datetime format (eg. Jan 2 1900 12:00 is a conversion of the number 1).
The result I'm trying to get as opposed to the above results is this:
+-----+------------+---------------------+
| ID | FieldName | Result |
+-----+------------+---------------------+
| 106 | Category | 1 |
| 107 | Species | 99 |
| 108 | Date | Jun 30 2017 12:00AM |
+-----+------------+---------------------+
Just for clarification, the Transaction table schema is as follows:
[ID] INT IDENTITY(1, 1) CONSTRAINT [PK_Transaction_ID] PRIMARY KEY,
[FieldName] VARCHAR(MAX) NULL
[Result] VARCHAR(MAX) NULL

Answer 1

SQL Server is making a guess at the data type for the CASE statement. It does this based on its internal precedence order for data types and the following case statement return type rule:
the highest precedence type from the set of types in
result_expressions and the optional else_result_expression.
Since int has a lower precedence order than datetime SQL Server is choosing to use a datetime return type.
Ultimately explicitly normalizing the data types of your case statement to varchar will solve the issue:
CASE WHEN #IDColumn = 1 THEN '1'
WHEN #IDColumn = 2 THEN '99'
WHEN #IDColumn = 3 THEN FORMAT([data].[ArrivalDate]), 'Mon d yyyy h:mmtt')
ELSE NULL
END
In case you are interested SQL Server uses the following precedence order for data types:
user-defined data types (highest)
sql_variant
xml
datetimeoffset
datetime2
datetime
smalldatetime
date
time
float
real
decimal
money
smallmoney
bigint
int
smallint
tinyint
bit
ntext
text
image
timestamp
uniqueidentifier
nvarchar (including nvarchar(max) )
nchar
varchar (including varchar(max) )
char
varbinary (including varbinary(max) )
binary (lowest)

Answer 2

It is converting the format because all types in a CASE should have the same format. I think you want to convert the date as a string (and the numbers too).
SELECT
,(SELECT Name FROM FieldNames WHERE ID=#IDColumn)
,CASE WHEN #IDColumn=1 THEN '1' --Result insert for FieldName 'Category' where ID=1 refers to 'Category Fruits'
WHEN #IDColumn=2 THEN '99' --Result insert for FieldName 'Species' where ID=99 refers to 'Apple - Fuji'
WHEN #IDColumn=3 THEN convert(varchar(MAX), [data].[ArrivalDate], 23) --Result insert for Field Name 'Date'
ELSE null
END
FROM #fruitTransactionData [data]

Answer 3

This is too long for a comment.
What you are trying to do just doesn't make sense. The columns you are inserting into are defined by:
INSERT INTO [dbo].[Transactions]([FieldName], [Result])
---------------------------------^ -----------^
The INSERT is inserting rows with two values, one for "FieldName" and the other for "Result".
So the SELECT portion should return two columns, no more, no fewer. Your SELECT appears to have four. Admittedly, the first two are syntactically incorrect CASE expressions, so the count might be off.
It is totally unclear to me what you want to do, so I can't make a more positive suggestion.