DEV Community

Cover image for Code Smell 234 - Long Circuit
Maxi Contieri
Maxi Contieri

Posted on • Originally published at maximilianocontieri.com

Code Smell 234 - Long Circuit

Be smart (and lazy) with low performant conditions

TL;DR: Premature Optimization is Evil. Optimization is Good.

Problems

  • Low Performance

Solutions

  1. Sort the conditions from faster to slower

Context

Readability is always essential and you should avoid premature optimization.

Non-premature optimization happens when you have actual evidence you can improve your code execution time without much readability penalizations.

Sample Code

Wrong

def is_warm():
    # This is a fast api call to your thermometer
    response = requests.get
        ("https://iot-device-api.example.com/current_temperature")
    temperature_data = response.json()

    return temperature_data.get('temperature', 0) > 25  

def is_weekend():
    # This function checks if today is a weekend based on a slow calendar API call
    response = requests.get
        ("https://calendar-api.example.com/today")
    calendar_data = response.json()

    return calendar_data.get('day_of_week', '').lower() 
        in ['saturday', 'sunday']

def is_sunny():
    # Very slow function to a low performant weather API call
    response = requests.get
        ("https://weather-api.example.com/current")
    weather_data = response.json()

    return weather_data.get('weather', '') == 'sunny'

is_sunny_value = is_sunny()
is_warm_value = is_warm()
is_weekend_value = is_weekend()  

if is_sunny_value and is_warm_value and is_weekend_value:
    # the 3 conditions are always evaluated
    print("Let's go outside!")
else:
    print("Stay at home.")
Enter fullscreen mode Exit fullscreen mode

Right

if is_warm() and is_weekend() and is_sunny():
    # the 3 conditions are evaluated in short circuit 
    # and sorted from fastest to slowest
    # for a fast exit
    print("Let's go outside!")
else:
    print("Stay at home.")
Enter fullscreen mode Exit fullscreen mode

Detection

[X] Semi-Automatic

You can detect slow calls using actual benchmarks.

Do not consider algorithm complexity since sometimes it is unrelated to actual data distribution. (for example, optimizing an array with a few elements).

Tags

  • Performance

Conclusion

Find bottlenecks using Pareto rules.

Optimize your code-critical sections.

Relations

Disclaimer

Code Smells are my opinion.

Credits

Photo by Nick Abrams on Unsplash


The key to performance is elegance, not battalions of special cases.

Jon Bentley and Douglas McIlroy


This article is part of the CodeSmell Series.

Top comments (0)