Learning Exercise
Errors happen. In Elixir, while people often say to "let it crash", there are times when we need to rescue the function call to a known good state to fulfill a software contract. In some languages, errors are used as a method of control flow, but in Elixir, this pattern is discouraged. We can often recognize functions that may raise an error just by their name: functions that raise errors are to have ! at the end of their name. This is in comparison with functions that return {:ok, value} or :error. Look at these library examples:
Map.fetch(%{a: 1}, :b)
# => :error
Map.fetch!(%{a: 1}, :b)
# => raises KeyError
Elixir provides a construct for rescuing from errors using try .. rescue
[]: # (elixir-formatter-disable-next-block)
try do #1
raise RuntimeError, "error" #2
rescue
e in RuntimeError -> :error #3
end
Let's examine this construct:
try.raise/2.rescue section, we pattern match on the Module name of the error raised
->:
e is matched to the error struct.in is a keyword.RuntimeError is the error that we want to rescue.
_ instead of the module name or omit the in keyword entirely.Errors (sometimes also called "exceptions") that you rescue this way are structs. Rescuing errors in Elixir is done very rarely. Usually the rescued error is logged or sent to an external monitoring service, and then reraised. This means we usually don't care about the internal structure of the specific error struct.
In the Exceptions concept you will learn more about error structs, including how to define your own custom error.
While working at Instruments of Texas, you are tasked to work on an experimental Reverse Polish Notation [RPN] calculator written in Elixir. Your team is having a problem with some operations raising errors and crashing the process. You have been tasked to write a function which wraps the operation function so that the errors can be handled more elegantly with idiomatic Elixir code.
Implement the function calculate!/2 to call the operation function with the stack as the only argument. The operation function is defined elsewhere, but you know that it can either complete successfully or raise an error.
stack = []
operation = fn _ -> :ok end
RPNCalculator.calculate!(stack, operation)
# => :ok
stack = []
operation = fn _ -> raise ArgumentError, "An error occurred" end
RPNCalculator.calculate!(stack, operation)
# => ** (ArgumentError) An error occurred
Function names that end in
!are a warning to programmers that this function may raise an error
When doing more research you notice that many functions use atoms and tuples to indicate their success/failure. Implement calculate/2 using this strategy.
stack = []
operation = fn _ -> "operation completed" end
RPNCalculator.calculate(stack, operation)
# => {:ok, "operation completed"}
stack = []
operation = fn _ -> raise ArgumentError, "An error occurred" end
RPNCalculator.calculate(stack, operation)
# => :error
Some of the errors contain important information that your coworkers need to have to ensure the correct operation of the system. Implement calculate_verbose/2 to pass on the error message. The error is a struct that has a :message field.
stack = []
operation = fn _ -> "operation completed" end
RPNCalculator.calculate_verbose(stack, operation)
# => {:ok, "operation completed"}
stack = []
operation = fn _ -> raise ArgumentError, "An error occurred" end
RPNCalculator.calculate_verbose(stack, operation)
# => {:error, "An error occurred"}
Edit via GitHub
Sign up to Exercism to learn and master Elixir with 57 concepts, 158 exercises, and real human mentoring, all for free.