Writing assertive code with Elixir

The post argues that assertive code in Elixir is clearer, faster, and easier to maintain. It recommends making assumptions explicit so unexpected input fails early instead of creating hidden complexity.

Reading notes

#

• Pattern matching is the preferred way to extract the token from a query-like string, because it asserts that each split pair has exactly two elements.
• The less strict version can accept malformed values by accident, such as a token value with extra equals signs.
• Using pattern matching turns unexpected shapes into crashes, which makes corner cases visible and easier to discuss.
• Polymorphism in Elixir is opt-in through protocols, so a function should only be opened to many data types when that is truly intended.
• Calling to_string before replacement makes dasherize work with many more types and therefore less specific.
• If the function should support only atoms and strings, the post recommends handling those cases directly instead of opening it to all protocol implementations.
• When a protocol is really desired, tests should cover at least a couple of supported types.
• For maps, strict field access is preferred over dynamic access because it catches bugs early.
• The same idea applies to structs, where user.first_name is preferred over user[:first_name].
• Deriving Access for a struct should be reserved for cases where dynamic access is truly needed.
• The overall claim is that assertive style postpones incidental or accidental complexity until it is actually required.