"MongoDB Code Smells: Defining, Classifying and Detecting Code Smells for MongoDB Interactions in Java Programs"

Défense de mémoire de messsieurs Kitzinger Thomas et Bernard Jehan

Rejoindre la défense

The code smells research and detection is an important topic in software maintenance and quality assessment. Indeed, since the introduction of the term by Martin Fowler and Kent Beck in 1999, it has been widely adopted and a lot of research about code smells has been conducted. NoSQL (Not only SQL) oriented database management systems appeared about ten years ago and now begin to be a subject of interest in scientific studies. Because of the emergence of new database management systems, new types of code smells need to be studied so that their persistence in these new systems can be avoided. This thesis aims at presenting the techniques we have defined and implemented to detect various code smells in the interactions between a Java program and a MongoDB database. We first defined a catalog to group and classify the code smells we could find in the literature. Then, we developed methods using CodeQL, a static code analysis tool, to detect instances of certain code smells that we had chosen in our catalog beforehand

Keywords
Static code analysis, Code smells, MongoDB, Taxonomy, Detection, Anti-patterns, NoSQL, CodeQL, Java

Télecharger : vCal