#
# Numeric-only value.
# The actual type should be inferred from the available values, as described below.
# If none of the values below are present, the value is considered empty.
#

int64[<=1] value_int            # Preferred over float if ambiguous
float32[<=1] value_float        # Only if int is empty