The Largest Case Study of Code Reviews—Ever

[article]

clearly in favor of the optimistic interpretation. There were many
cases where reviewers argued with the preparatory comments and where reviewers
clearly were looking at other parts of the code and thinking about ramifications
elsewhere in the code base.

So, our conclusion is that author preparation is indeed a Good Thing and that it saves time overall during the review because reviewers are not having to call out "obvious" defects.

Results in summary
To summarize all our results, including some things not discussed here:

    • Lightweight-style reviews are effective and efficient
    • Review fewer than 200-400 LOC at a time
    • Aim for an inspection rate of less than 300-500 LOC/hour
    • Take enough time for a proper, slow review, but not more than 60-90 minutes
    • Author preparation is good
    • Expect defect rates around 15/hour, higher only when <150 LOC under review
    • Left to their own devices, reviewers' inspection rates will vary widely,  even with similar authors, reviewers, files, and review size

For a full explanation and evidence for all these results, refer to  Best Kept Secrets of Peer Code Review .

Cisco® and MeetingPlace® are registered trademarks of Cisco Systems Inc.. These names and the information herein are reproduced with permission.

About the author

AgileConnection is one of the growing communities of the TechWell network.

Featuring fresh, insightful stories, TechWell.com is the place to go for what is happening in software development and delivery.  Join the conversation now!