0
I was facing problems to carry out the maintenance in a code of my own, due to the extensive and little defined classes (grew a lot over time), to simplify the question I will divide the time line of the code in version 1 and 2.
Version 1:
Practically all screen construction and variable initialization occurred in the main method, although we had a relatively large number of external classes (Components, entities etc.) and the classes related to the database were divided into Select, Delete, Update and Insert.
As reported above the main difficulty here was maintenance, although it was not so difficult to find the methods ended up getting too much code unused over time and the propagation of errors was difficult to deal with.
Version 2:
Almost the entire code has been redistributed, splitting packages by functionality and breaking code into smaller classes with more defined purpose.
Immediately the maintenance of the code became much easier besides allowing me to extend functionalities with almost zero propagation of errors. However the performance of the application dropped noticeably.
During university, I believe that because we produce small projects and personal use, we don’t have to worry too much about the relationship between performance and scalability.
My question is whether, "in the real world," these two metrics are irreconcilable, so that investing in a necessarily we’ll have to give up the other?
Note: For purists, let’s adopt the application response time as a metric for performance.
When it comes to code organization, I’ve never seen performance loss being discussed. The suspicion falls on some piece of code or programming error that is causing this. It would be the case to investigate, looking for tools that help in this process of investigation.
– Piovezan
I also have this suspicion, but I took the situation to check if in practice we have to really balance performance and scalability.
– Gustavo Fragoso