Airtable formulas interview

1. Have to define what formulas look like
   • How are formulas (specifically operators) defined
     • How are we delineating column names from other types of objects in the formula
     • ([A] + [B])
     • =SUM(A, B)
     • Whether we will allow both kinds of operators (prefix vs infix)
       • Design decision: allow both kinds of operators
   • What kinds of inputs/outputs can formulas have
     • Do we want to enforce types (raise an error if you’re applying an operation on forbidden types, or fail silently)
2. How to actually evaluate formulas
   • Storing some internal mapping from operator in names in Airtable to the programmatic operator being applied
     • “SUM” -> operator.add
   • “=SUM([A], 2)”
     • Parsing step:
       1. Extract all function compositions with functions and corresponding inputs from the formula string

       class Function:
           def 
           operator: # type: Callable
           Inputs: List[Union[Function, Literal]]

       Function(sum, [Function(multiply, 3, 3) , 2])

       Def eval(input: Input): If input is Literal: Return input For input in inputs: evaluated_input = eval()
     • How will we extract useful information from this formula string
       • Extracting the function name
       • Detecting whether we’re referencing an existing column, or just using a literal
         • Substituting value from existing column before you can proceed with the rest of the computation

Formula for Column A depends on Column B, and A also references C

Circular dependency - Column A depends on Column B, and formulas for Column B depends on Column A

Formula A: Sum(1, 2) + Sum(B, C)

Sum(Sum(1, 2), Sum(B, C))

• At some point in the evaluation (maybe at the innermost layer), we need to substitute in values from other columns

• ---

  uid: 202011181541 tags: #airtable #interviews

---

---

---