Profiling

Enables quick measurement of function execution time without modifying the function's internal code.

Web backend: measuring request handler latency during load testing Data pipeline: timing a data transformation step to identify bottlenecks Scientific computing: benchmarking a numerical kernel function.

Python applications, performance testing scripts, scientific computing libraries, web frameworks like Flask or Django.

@timer

@timer

• functools.wraps to preserve function metadata
• time.perf_counter for high-resolution timing
• logging module to output results.

• manual timing with time.time() inside function (more intrusive)
• using cProfile module for detailed profiling (heavier overhead)
• using line_profiler for line-level details (requires installation).

Forgot to include parentheses when decorator takes arguments (causing TypeError) Applied @timer to a class method without proper binding (leading to missing self argument) Forgot to return the wrapper function (causing decorated function to return None) Applied decorator to a generator function without preserving yield behavior (breaking iteration) Used mutable default arguments in decorator factory (leading to shared state across calls)

@profile (from memory_profiler): measures memory usage instead of time @lru_cache: caches function results based on arguments contextlib.timer (hypothetical): uses with statement instead of decorator

Coming from Java: may expect annotation syntax like @Timed to modify method behavior at runtime — Python decorators actually wrap the function returning a new callable.

• @profile
• @lru_cache
• @staticmethod

Avoid using @timer in production code paths where overhead matters, as even minimal wrapping adds function call overhead; the decorator adds a small constant overhead per call (typically a few microseconds), which can accumulate in tight loops; be careful when decorating recursive functions, as each recursive call will be timed individually, potentially skewing results and increasing overhead exponentially.

Think of @timer as a stopwatch you clip onto a function — every time the function runs, the stopwatch starts and stops automatically, telling you how long it took.

Use cProfile for function-level profiling or line_profiler for line-level details.