The basic idea behind this question is to gather information related to types of faults that are specific to code reading or code review.
Those techniques are usually used for finding defects related to the internal quality of software (ISO 9126). Example would be a missing architectural link between two components.
Suggested reading:
Mantyla, M.V.; Lassenius, C. "What Types of Defects Are Really Discovered in Code Reviews?". IEEE Transactions on Software Engineering. May–June 2009
Its not only the fact to detect defects by code reading and reviewing. As Peter mentioned first the syntax and then the semantics of code.
But code is also part of communication. If an other programmer first can read and understand written code by another programmer than the code is maintainable. Also the application of known pattern can be checked by a review.
Conclusion:
Reviews support the detection of defects, improve the readbility and maintainability of code and support the application of well known and understood programming patterns
From the commercial world, code reviews are only about read-ability and allowing teams to learn the code as a group from the author. The challenge is that you typically do not have key knowledge holders (architecture team members) to participate in all of the reviews that the execution team manages.
There are significant tools that look at errors in syntax, language usage, missing architectural links, etc. But, historical data from numerous companies have indicated that the people aspect of the process is not where you would hope it would be. In long code reviews, the reviews do not typically find logic errors (depending on the complexity of the code and architecture).
Instead of these approaches, there is the want to pursue a Test Driven Design (TDD) and paired programming to address the concerns I listed above. But, this is from a commercial world view where time, resources, and business demands require tradeoffs.
Code review helps in identifying the coverage of code that has been developed. for example if there are multiple functions defined for a component and if the call to a function is missed with in the component, it can be identified in the code review. similarly code review helps to identify if all the required loops and lines are being used by the application or not. It helps in identifying issues and fix them in early stages which impacts the cost of building a software in a good way.
Note that ISO/IEC 9126 now has been abandoned and replaced by ISO/IEC 25010 and related standards in the 250xx range.
As per my discussion with the data scientist the codes which are built now a days are automatically tested when they write the code and they are suggesting syntactic and semantic errors are having more possibilities to end up in defect.
This link talk about:
EFFECTIVE CODE REVIEW: Uprooting Source Code Defects
https://codingsight.com/effective-code-review-uprooting-source-code-defects/
- Badges
- Science topic
- Similar topics
- Human-Computer Interaction
- Collaboration