When not to use OOP in Python

Python has the flexibility to develop programs using either OOP languages such as Java or using declarative programming such as C. OOP is always appealing to developers because it provides powerful tools such as encapsulation, abstraction, inheritance, and polymorphism, but these tools may not fit every scenario and use case. These tools are more beneficial when used to build a large and complex application, especially one that involves user interfaces (UIs) and user interactions.

If your program is more like a script that has to execute certain tasks and there is no need to keep the state of objects, using OOP is overkill. Data science applications and intensive data processing are examples where it is less important to use OOP but more important to define how to execute tasks in a certain order to achieve goals. A real-world example is writing client programs for executing data-intensive jobs on a cluster of nodes, such as Apache Spark for parallel processing. Here are a few more scenarios where using OOP is not necessary:

• Reading a file, applying logic, and writing back to a new file is a type of program that is easier to implement using functions in a module rather than using OOP.

• Configuring devices using Python is very popular and it is another candidate to be done using regular functions.

• Parsing and transforming data from one format to another format is also a use case that can be programmed by using declarative programming rather than OOP.

• Porting an old code base to a new one with OOP is not a good idea. We need to remember that the old code may not be built using OOP design patterns and we may end up with non-OOP functions wrapped in classes and objects that are hard to maintain and extend.

In short, it is important to analyze the problem statement and requirements first before choosing whether to use OOP or not. It also depends on which third-party libraries you will be using with your program. If you are required to extend classes from third-party libraries, you will have to go along with OOP in that case.