Code manipulates strings and hashes directly instead of using domain objects.
protocol = url.match(/^([^:]+):/)[1]username = user["name"]- Since you can't define new methods on primitive data types (at least not without causing a great deal of confusion), logic for manipulating them gets scattered around the codebase.
- Your code becomes coupled to the underlying structure of the data, making it harder to test and reason about.
- You have to think about error cases more often. For example, when using a string to represent a URL, you have to think about the case where the string is not a well-formed URL every time you do something with it. For more on this idea, see Alexis King's post "Parse, Don't Validate".