Defect management methods can be divided into defect prevention and defect exclude from the perspective of management timing. Defect prevention achieves the purpose of defect management by using a formal method, multiplexing, mode, or good programming method. The management of existing defects mainly uses the following two groups of methods: direct defect identification method and failure detection and root cause analysis method.
1. The defect management method of defect identification methods can generally be referred to as static verification, peer review or software analysis. These methods are directly, static, detailed research-based, detailed research-based research on products (eg, source code), based on software performance, to demonstrate failure (such as test). Static verification can be applied to all software process products, which makes we can detect and eliminate defects in early detection and exclude defects (such as design review).
In general, the terminology in the defect identification method is easy to confuse, because the name of the method usually does not correctly or sufficiently describing a given method, the same term, can refer to a variety of different methods, the same method may There are multiple different names. To this end, the definitions proposed herein are discussed here, in which the classification of the defect management method is based on the intent or main goal of each method. The main objectives of this classification method are: assessment, training, and optional inspection and defect identification. This type of classification can not only lay the foundation for communication, but also help compare different methods in this classification.
1) Assessment
The main purpose of many static methods is to assess project progress, resource allocation, adequacy, and whether it is consistent with the agreement. This method is called a review (REVIEW). The methods described in this classification include technical review and management reviews. The purpose of the technical review is to assess the consistency of the statute and project plan to ensure the integrity of the change process. Management reviews are not to detect and exclude defects as the primary goal, but to determine that project progress is introduced into corrective actions and ensure appropriate allocation of resources. Because there is no clear number of members of the assessment team, there is no precise definition process, process objectives and its delivery products, so it is difficult to measure the efficacy of review technology.
2) Training and optional options
Take the exam belong to this type. Walk-through is the leadership of product creators, a semi-typed review of products by numerous statistics. This review tries to meet different goals, including participant training, optional option, readability and module check, defect logo. The investigation is considered an effective way because: can evaluate a large number of products in a relatively short time; the team members do not need too much special training; if the team does not need to consume Many time and effort; in addition, there is a variety of goals that need to meet, usually lack of formal process models, lack of data collection requirements or change cooperation has made it a useful way. But relative to the review, the effectiveness is poor.
3) Defect logo
Many methods identify defects as their main goal. These methods are referred to as an Inspection. There are a lot of introduction to the review method, which is basically the formation of the regulation of Fagan1976. Due to the considerable importance of the review in defect management, it is recommended here that Strauss and Ebenau style review, that is, the process is reviewed. The basic principles of the review were the same as those originally proposed by Fagan, but the angle has improved because the review is used to enhance the quality of the products such as hardware, documentation, courseware and more. This approach emphasizes the following aspects: early applications for review, apply to a wider range of products, can customize to meet specific project environments and specific product types.
The review to the software is currently only the most popular defect management method of unit testing, which is the main method of static defect identification. All experiences on review are positive. Some of the most important benefits of using the review method in obtaining defect management and quality data, increasing the possibility of completing the task before the final deadline, has been improved, and the morale of the person has also improved.
2. Defects for failure detection is an activity that uses a software environment, which is intended to identify, indicating that the results are deviated from expected. This method can be broken down into three different activities: failure detection, root cause analysis and root cause defect removal. Failure detection is an activity that is closest to traditional testing from the sense of defect exclusion. Despite certain differences, the difference between these two terms is not large, and sometimes it can be interchanged.
The root cause analysis is an activity of defects or a series of defects (fail) to detect the defects in the source code caused by failure detection or other common methods. This is a very time consuming task.
The root cause defect exclude is an activity that has been identified in the source code through the previous two activities. The two activities are commonly referred to as debugging.
3, the comparison and testing of review and testing is not a method of replacing each other. Both methods have their strengths, the best choice for managing specific types, such as defects in formats can be found, integrated The failure is best to be tested. At the same time, test cases and test documents may be defective, and this is improved by review.
The review can be applied early in the early days, which can be applied to the early product of lifecycle, and the test can only be applied when performing a complete code. This makes the review have greater efficacy than testing. This is because, fundamentally, the proportion of all defects found in a method and the stage of using this method are determined by this method. In addition, the review is always better than the test from the measurement of the percentage of the discovered defects and the time required to exclude defects. It can be said that at least in general, the review method is a very effective defect management method compared to the test. Despite this, it is also difficult to obtain a predictive model of the determined review efficacy from the absence of defects that can be excluded within the unit time.
