Min identified as aggregation function instead of transform function

[mneedham]

When I try to use the min scalar function e.g.

select min(distance, 5) 
from courses limit 10

It instead uses the aggregation function. i.e. it finds the minimum value across all rows, rather than computing the minimum value of distance and 5 per row

If I would use the function with some hardcoded values then it does pick the correct function:

select min(7, 5) 
from courses limit 10

The same issue is there for the max function as well.

[richardstartin]

min is an aggregation function in SQL. I think you're trying to use it like least (max: greatest) which isn't implemented in Pinot.

[mneedham]

Pinot has two different min functions. There's the aggregation function that you mention, but there is also a scalar function defined here - https://github.com/apache/pinot/blob/master/pinot-common/src/main/java/org/apache/pinot/common/function/scalar/ArithmeticFunctions.java#L57

  @ScalarFunction
  public static double min(double a, double b) {
    return Double.min(a, b);
  }

  @ScalarFunction
  public static double max(double a, double b) {
    return Double.max(a, b);
  }

So I wanted to call this min function in my query, but it calls the other one. So is your suggestion that the scalar functions should have different names?

[richardstartin]

Pinot has two different min functions. There's the aggregation function that you mention, but there is also a scalar function defined here - https://github.com/apache/pinot/blob/master/pinot-common/src/main/java/org/apache/pinot/common/function/scalar/ArithmeticFunctions.java#L57

  @ScalarFunction
  public static double min(double a, double b) {
    return Double.min(a, b);
  }

  @ScalarFunction
  public static double max(double a, double b) {
    return Double.max(a, b);
  }

So I wanted to call this min function in my query, but it calls the other one. So is your suggestion that the scalar functions should have different names?

I’m aware of the scalar arithmetic functions. In SQL this function, which operates on tuples, not sets, is called least, not min. min is an aggregation over a set. I experimented with adding least and greatest scalar functions, and it works, but I would prefer to implement this as a TransformFunction for efficiency reasons (vectorisation).

[mneedham]

Got it, thanks!