A problem with Primitive Obsession

By an odd coincidence, a number of the teams I work with have been looking at the Primitive Obsession code smell recently as part of the software habitability and craftsmanship training programmes I am currently running. As we have explored legacy code looking for examples of the smell, I’ve noticed a tendency to enthusiastically label every primitive in any method signature as smelly. This certainly demonstrates Primitive Obsession, but mostly on the part of the reader of the code, rather than in the code itself.

I so wish Fowler and Beck hadn’t used this particular name, because it seems to spread as much confusion as enlightenment. In my book I tried renaming it to Open Secret, but that was a very small voice lost in the noise. So hear my plea: please, please stop using the term Primitive Obsession — let’s talk about Connascence of Meaning instead.

I think it is important to recognise that not every primitive in a method signature represents (bad) Connascence of Meaning: Firstly, the problem is significantly weaker when the method is private (ie. it can only be called from inside the same object). That’s because connascence becomes stronger with distance. Inside a single class we expect a lot of connascence, because otherwise the class is probably not a cohesive unit of responsibility. It’s a fairly weak smell when two methods in a class both know how to interpret the meaning of a primitive.

And secondly, recall that this smell is mostly about the way domain concepts are represented and understood. If the primitive is passed straight through our code without being parsed in any way, again the smell is much weaker because we are not attempting to intuit any Meaning for ourselves. For example, if you’re working in a domain in which time is always given to you as a long, and you never do anything with it except store it or pass it on, then there is certainly Connascence of Meaning between the libraries you’re using, but I would argue that there’s probably none in your own code.

In general I’m finding that connascence is better than code smells as a tool for discussing coupling. And in particular, it seems to me that Primitive Obsession creates the wrong impression and makes the concept harder to learn.


One thought on “A problem with Primitive Obsession

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s