【文章內(nèi)容簡介】 effect graph. The graph is a digitallogic circuit, but in order to build a graph no knowledge of electronics is necessary. The tester should understand only the boolean logic. The following steps are used to produce test cases [2]: – Divide the specification into workable parts. Large specifications make a causeeffect 畢業(yè)設(shè)計(jì)（論文） 外文資料 原文 5 graph difficult to manage. Figure : Simple CauseEffect Graphs – Identify the causes and effects in the specification. A cause is a distinct input condition or an equivalence class of input conditions. An effect is an output condition or a system transformation. The causes and effects are identified by reading the specification. Once identified, each cause and effect is assigned an unique number. – From cause and effect information a boolean causeeffect graph that links causes and effects together is created. – Annotations with constraints are added, that describe binations of causes and/or effects which are impossible. – The graph is converted to a decision table. – The colomns of the decision table are converted into test cases. The simple examples of causeeffects graphs are shown in Figure . The more detailed description with examples of this method can be found in [1] and [2]. Error Guessing Design of test cases using error guessing method is based on the tester39。s past experience and intuition. It is impossible to give a procedure for an error guessing approach since it is more intuitive and ad hoc process. The basic idea behind is to enumerate a list of possible errors and then write test cases based on this list. State Transition Testing State Transition Testing can be used for both objectoriented and procedural software development. The approach is based on the concept of finitestate machine and states. It views the software 畢業(yè)設(shè)計(jì)（論文） 外文資料 原文 6 under test in term of its states, transitions between states, and the inputs or events that trigger state changes. A state is an internal configuration of a system. It is defined in terms of the values assumed at a particular time for the variables that characterize the system or ponent [1]. A finitestate machine is an abstract machine that can be represented by a state graph having a finite number of states and a finite number of transitions between states [1]. A State Transition Graph (STG) can be designed for the whole software system or for its specific modules. The STG graph consists of nodes (circles, ovals, rounded rectangles) that represent states and arrows between nodes that indicate what input (event) will cause a transition between two linked states. The Figure shows a simple state transition graph [1]. Figure : A simple state transition graph S1 and S2 are two states. The black dot is a pointer to an initial state from outside. The arrows represent inputs/actions that cause the state transformations. It is useful to attach to the graph the system variables that are effected by state changes. The state transition graph can bee very plex for large systems. One way to simplify it to use a state table representation. A state table for the graph in Figure is shown in Table 2 [1]. The State Table lists all inputs that cause the state transitions. For each state and each input the next state and action taken are shown. Table 2: A state table for the state transition graph in Fig. The STG should be prepared by developers as a part of the requirements specification. Once the graph was designed it must be reviewed. The review should ensure that – the proper number of states is represented。 畢業(yè)設(shè)計(jì)（論文） 外文資料 原文 7 – each state transition (input/output/action) is correct。 – equivalent states are identified。 – unreachable and dead states are identified. Unreachable states are states that will never be reached with any input sequence and may indicate missing transitions. Dead states are states that once entered can not be exited [1]. After the review the test cases should be planed. One practical approach is to test every possible state transition [4]. 3. Automated BlackBox Testing A few blackbox methods were listed above. The problem with those methods is that often the performance of testing depends greatly on experience and intuition of the tester. Therefore, there is a question if blackbox testing can be automated to make testing more thorough and costeffective. Furthermore, there is need in blackbox methods, that can b