Lists or Objects

"Our rule of thumb is that we try to limit passing around types with generics (the types closed in angle brackets). Particularly when applied to collections, we view it as a form of duplication. It's a hint that there's a domain concept that should be extracted into a type." [Growing Object-Oriented Software, Guided by Tests - Steve Freeman, Nat Pryce, page 136)]


Not ideal - business logic will leak.

Better - quotes are now encapsulated.


The benefit of the second approach is that you begin to force the collection to do the work. Consider a collection of quotes - using the first approach a developers' typical instinct would be to loop over the list and check each quote until the lowest quote was found. You could argue that this in turn violates the 'Tell Don't Ask' principle because very quickly you can find yourself digging down into objects. We would not do this with standalone objects, so collections should not be any different. Yet as developers we often avoid creating objects to encapsulate collections, despite this principle being taught in the early chapters of any OO programming beginners book.

Another key point of encapsulating collections is that you reduce duplication. If another part of the application needs to perform a check for the lowest quote, you'd end up either duplicating the code to loop over the quotes, or creating a helper method to do the actual processing. Using the collection means such features are only a method call away at all times.

In a recent code kata as well as a current set of work, we are applying some of the above concepts and finding a dramatic increase in the quality of our code. Those of you may have noticed that the above implementation inherits from a standard list, therefore it is indeed possible to still violate encapsulation and dig through the collection yourself. There are pros and cons to this approach, one benefit is all of the out of the box functionality you get for free - the ability to loop over the quotes (for displaying say) and return a count for example. The downside being developers can treat these collections as if they were a list.

In C# it is possible to use indexersor interfaces to improve the above implementation, though we tend to agree that it's up to the developers to do the "right" thing - in other words, not dive down into the collection. Interesting this very topic has been discussed on the excellent StackOverflow - I and others will sure agree that the answer(s) provided are well off the mark. Encapsulate your public collections - do not treat them as primitives!