This fallback mechanism was introduced when adding complexity_in_classes and complexity_in_functions measures, in order to give some time for language plugins to add these measures.
Here's how is computed the complexity per class (and it's the same for the complexity per function) :
- Take the complexity in classes (complexity_in_classes)
- And divide by the number of class
If complexity in classes is not available, then we use the complexity. But as this can lead to some inconsistency (how do you know from where this complexity is coming ? Classes or functions ?), we should remove this fallback mechanism for these measures.