[BUG] ROUND function returns FP result for FP value

[Yury-Fridlyand]

What is the bug?

sql/core/src/main/java/org/opensearch/sql/expression/operator/arthmetic/MathematicalFunction.java

Lines 411 to 419 in 43ceda1

	FunctionDSL.impl(
	FunctionDSL.nullMissingHandling(
	v -> new ExprDoubleValue((double) Math.round(v.floatValue()))),
	DOUBLE, FLOAT),
	FunctionDSL.impl(
	FunctionDSL.nullMissingHandling(
	v -> new ExprDoubleValue(new BigDecimal(v.doubleValue()).setScale(0,
	RoundingMode.HALF_UP).doubleValue())),
	DOUBLE, DOUBLE),

sql/core/src/main/java/org/opensearch/sql/expression/operator/arthmetic/MathematicalFunction.java

Lines 397 to 398 in 43ceda1

	* (x: FLOAT [, y: INTEGER]) -> FLOAT
	* (x: DOUBLE [, y: INTEGER]) -> DOUBLE

How can one reproduce the bug?

opensearchsql> select typeof(round(1.1)), round(1.1);
fetched rows / total rows = 1/1
+----------------------+--------------+
| typeof(round(1.1))   | round(1.1)   |
|----------------------+--------------|
| DOUBLE               | 1.0          |
+----------------------+--------------+

What is the expected behavior?
ROUND function should return an integer value (INTEGER or LONG).

It may happen that returned FP value is not an integer number (has digits after the point).

[matthewryanwells]

@Yury-Fridlyand Currently if you put a double in you get a double out, and if you put a long in you get a long out.

If you put a double into round(x) should it return a long always? If you put a double into round(x, y) you can specify how many decimal points to return so sometimes it would have decimals and sometimes it wouldn't, should this always return a double?

[Yury-Fridlyand]

I expected round(double x) to return long and java's Math.round does. On another hand, MySQL returns double in that case, and we're aligning with MySQL.
Probably I got confused between these two behaviours and it is not a bug actually.