Hello all,

This PR implements the CreateVectorizedFunction using only template parameters, it's a complement to the UDF C++ API.

It's very similar to the first CreateFunction implemented by the PR #712, the main difference is in the second argument that instead of receiving a generic function pointer (i.e., TR (*udf_func)(Args…)), it receives a vectorized function pointer of the type scalar_function_t:

typedef std::function<void(DataChunk &, ExpressionState &, Vector &)> scalar_function_t;

1. template<typename TR, typename... Args> void CreateVectorizedFunction(string name, scalar_function_t udf_func)

• template parameters:
  • TR is the return type of the UDF function;
  • Args are the arguments up to 3 for the UDF function.
• name: is the name to register the UDF function;
• udf_func: is a vectorized UDF function.

This function automatically discovers from the template typenames the corresponding SQLTypes:

• bool → SQLType::BOOLEAN;
• int8_t →SQLType::TINYINT;
• int16_t →SQLType::SMALLINT
• int32_t →SQLType::INTEGER
• int64_t →SQLType::BIGINT
• float →SQLType::FLOAT
• double →SQLType::DOUBLE
• string_t →SQLType::VARCHAR

An example of use would be:

static void udf_unary_function(DataChunk &input, ExpressionState &state, Vector &result) {
	assert(input.column_count() == 1);
	assert(input.data[0].type == TypeId::INTEGER);

	result.vector_type = VectorType::FLAT_VECTOR;
	auto result_data = FlatVector::GetData<int>(result);
	auto ldata = FlatVector::GetData<int>(input.data[0]);

	FlatVector::SetNullmask(result, FlatVector::Nullmask(input.data[0]));
	for (idx_t i = 0; i < input.size(); i++) {
		result_data[i] = ldata[i];
	}
}

connection.CreateVectorizedFunction<int, int>("udf_unary_int_function", &udf_unary_function);

P.S. It's missing a CreateVectorizedFunction to disambiguate some SQL types (some primitive types, e.g., int32_t, are mapped to the same SQLType: INTEGER, TIME and DATE). Such function will be implemented soon in the same style of the second CreateFunction present in the PR #712, with some extra SQLType arguments.