The goal of the X model is to make up for some of the defects of the V model. X The model really solves all aspects of the test process, such as handover, often integration? In software testing, the V model is the most widely known model, although many practical testors are still less familiar with V models, or other models. The V model has existed for a long time, and the waterfall development model has a common feature, which is also criticized and questioned with the waterfall model. The V model has been discussed in detail in "Software Test: Unnearable Stage", which is only a summary introduction. The process in the V model is from left to right, describing basic development processes and test behavior. The value of the V model is that it clearly indicates the different levels of the test process and clearly describes the correspondence between these test phases and the development processes.
Figure 1 - V Model diagram
In the V model, the unit test is based on the code-based test, initially executed by the developer to verify that the various parts of its executable program code have reached the expected functional requirements; the integrated test verifies between 2 or more units Whether the integration is correct, and has a targeted interface between the units defined in the detailed design; after all unit tests and integrated tests are completed, the system test begins with the operation of the customer environment simulation system to verify the system. Whether it has reached the functions and performance defined in the profession; Finally, after the technical department has completed all the tests, it is accepted by the business expert or the user to ensure that the product can actually meet the needs of the user business. Although many people have denied V models, few people really discuss these issues in detail. Brian Marken ("The Craft of Software, 1995)" Author) "Author) is said. At STAR2000 (Software Testing Analysis and Review) Eastern Meeting, March has conducted an argument with Dorothy Graham (another author of this series of articles), and presented the V model on its personal website www.testing.com Some pertinent observations.
X model
Marke has some views and comments, of which first is Marken's not recommended to establish an alternative model. Here I take into liberty to reference some of the March ideas and reheated the organization, forming "X model." In fact, it is not to choose such a name for the V-model, but because of the other reasons: X usually represents unknown, and March also believes that his view is not enough to support a full description of a model, but there is a model. Some of the main contents, which also include highlights such as Explorative Testing. I also need to apologize to march in terms of using text, because I'm in question, most of them belong to X generation (X generation). In addition, I taught a diagram of a X shape. I believe that this figure can reflect the view of Marken in another expression.
Figure 2 - X Model Schematic
Since the X model has never been documentized, its contents need to be inferred from the related content of the V model, which is embodied in the relevant article of Marick. The discussion of the X model is relatively short, because it has not been fully expanded from the text from the text, and I don't want to repeat the "Software Test: Inloaded Stage" article in the "Software Test: Unbuterated" article. Many test technologies mentioned in the article. concept. The most important criticism of Marken's V model is that the V model cannot boot all the processes. He believes that a model must be able to handle all aspects of development, including integration, frequent repetition, and lack of demand documents, and so on.
Solve the issue of handover and frequently integrated cycles
March believes that a model should not specify those behaviors that are inconsistent with the currently recognized practice. I also agree with this. The left side of the x model is described in the separate encoding and testing of separate program segments, which will then communicate frequently, through integration, ultimately synthesized as executable programs. (The upper right half), these executables also need to test. The finished product of the integrated test can be sealed and submitted to the user, and can be used as part of the integrated integration within a larger scale and range. Multi-parallel curves indicate that changes can occur in each part. As can be seen from the above figure, the X model is also positioned with exploratory test (bottom right). This is a special type of test that does not conduct prior programs, such as "What will I do this?", This way can help experience more software errors outside the test plan. Although Marke did not clearly explain this, it must be very happy to see the definition of the method. However, focusing on such low-level behaviors may cause different discussion. A model is different from a separate project plan. The model should not describe the specific details of each project, and the model should guide and support the project. Of course, the handover of the code can also be simply considered to be an integrated form. The V model does not limit the number of occurrences of various creation cycles. Both March and Graham agree that test design should be performed before the test is performed. March: "Design when you master the relevant knowledge, test when your hand is delivered," The X model contains the steps of test design, just like the steps you want to include using different test tools, and the V model does not Do this. However, Marick's example prompts that the X model is not a real model in this language. Instead, it should be allowed to choose the test design step at any time. Prior plan
March questioned the V model, and because the V model is based on a set of development steps that must be strictly arranged in a certain order, this is likely to not reflect the actual practice process. Although many projects lack sufficient demand, the V model starts from demand processing. V Model Tips We have to test the content you have received during the development phase, but it doesn't specify how much we have to get. If there is no need information, the developer knows what they want? I advocate that there is enough demand in the X model and other models and at least once. Although there must be normal work in the case of no model, an effective model can encourage many good practical methods. Therefore, a strong term of the V model is its confirmation of its clear demand role, and the X model does not do this, which is probably a shortcoming of the X model. March also questioned the difference between unit testing and integration tests, because people may skip unit testing in some occasions, and are passionate about integration testing directly. March worried that people were blindly followed by the "V-model of the school", according to the steps guided by the model, and in fact certain practices did not fit practical. I have made my efforts to combine Marken's expectations to combine the expectations of behaviors that need scalable behavior, so that the X model does not require an integral part of the creation of executable (top of the figure). Before integrated tests, unit tests are performed on each program (the behavior on the left side of the figure). However, the X model has not provided whether to skip the judgment criterion of unit testing.
Outside of "Stage"
