Optimization is normally considered a good practice. However, this doesn't hold true when the act of optimization ends up driving the design decisions of the software solution.

A very common pitfall developers face while starting to code a new piece of software is premature optimization.

When this happens, the end result ends up being quite the opposite of the intended optimized code. It can contain an incomplete version of the required solution, or it can even contain errors derived from the optimization-driven design decisions.

As a normal rule of thumb, if you haven't measured (profiled) your code, optimizing it might not be the best idea. First, focus on readable code. Then, profile it and find out where the real bottlenecks are, and as a final step, perform the actual optimization.