Table 25 Summary of the main findings
Factor | Findings |
|---|---|
Design comprehension | A design comprehension tool support does not improve agreement on god class detection |
Developer experience | The experience impacts agreement on god class detection, but not definitively |
Developer knowledge | It should be further analyzed |
Software size | Software size impacts agreement on god class detection |
Developer training | Participants discussing the conceptualization and personal thresholds based on examples agree and succeed more than participants in a reading-based training |
Overall discussion | God class detection is more related to human traits than to an overall understanding of the code design |
Other context factors not related to the human traits, such as software size, also affect human conceptualization of smells. This issue should be further investigated | |
New heuristics for smell detection should be presented after the evaluation of the relationship between two aspects: 1) the current heuristics and 2) the context in which the smells are used (human traits there included) | |
The area should addresses the code smell conceptualization problem, which involves the discussion about ensuring same comprehension of the smell concepts and similar personal criteria or thresholds |