SDO Best Practices Catalog - Code Review Process

Benefits

• The strongly defined code review process allows to avoid functional and non-functional defects appearing, prevents anti-patterns and bad practices in the code, increases maintenance of the product, forces code conventions across the team, and lets the team members share knowledge between each other.

• The code review increases responsibility for the code quality by each team member.

• Using a code review tool simplifies the code review, when it is needed to involve several reviewers, enables involvement of an observer role, and decouples an author and a reviewer, as the author is not required to wait for the reviewer to start a review. In addition, the reviewer is able to review just updated changes in comparison to failed review changes that simplifies and speeds up the review process.

Indicators of Successful Process

• The code review procedure is defined and documented

• Code review checklist is used

• Code is reviewed against requirements (code and project conventions, specified architecture, functional and non-functional requirements, security guidelines)

• Code review is used as learning and knowledge sharing instrument and share typical findings among team members

• Code review workflow is using best practices for selected tool

• Code review is automated (specialized review tools are used)

• All changes are reviewed and validated

• Consistent code conventions are enforced through all product parts or modules

• Code review process roles are assigned: referee, author, reviewer. One person cannot have more than one role.

• Code review is done in small portions: up to 400 lines per review, up to 2 hours per day for reviewer and collect metrics to enforce that practice

• A wiki of project best practices are up to date and those practices are enforced on review

• Reviews are optimized:

• • • Drop trivial changes review (e. g. some autogenerated files, etc.)

    • Setup focused review sessions guided by author to speedup (author comments changes and provides context for made decisions)

    • Involve only necessary reviewers

    • Author critically reviews own code before requesting a code review

    • Review statistics is collected and used to optimize review process

• Process ensures that identified code review defects or action items are fixed

• Feedback cycles are reduced by introducing relevant, intelligent comments

• Code review results are reviewed and approved by project leadership team

• Code review items are closed along with the story and are part of acceptance criteria