【正文】 in Figure 2 (a), none of the statements can be split while maintaining the data dependences existing in the servlet program shown in Figure 1 (b). However, if the data dependences for the variable out are ignored, the data dependence graph is changed as shown in Figure 2 (b). Using the data dependence graph shown in Figure 2 (b) makes it possible to split S2 from the set of S9, S10, S11, and S12 which cannot be split due to the data dependences for the variable book. In servlet programs piled by the Apache Jasper piler, the data dependences for the variables shown in Table1 can be ignored for the same reason as for the variable the values of these variables are defined at the top of each servlet program and are frequently referred to, most statements cannot be split if the data dependences are to be maintained. In general, the outoforder execution is allowed for the statements without data and control dependences. Ignoring the data dependences for the variables shown in Table 1 may introduce an incorrect execution order of statements. For example,if S10 and S12 shown in Figure 1 are split into a JSP fragment and the JSP fragment is included before S11, an incorrect Web page is generated. In order to avoid such incorrect execution, our algorithm does not allow changes in the order of the statements that have ignored data dependences. . Analysis of Data Dependences for Java beans In a JSP page, Java beans are frequently used to store and retrieve data as property values. Each Java bean is a Java object and developers can use Java beans by using three kinds of tags in a JSP page. ● jsp:useBean: Declares a named Java bean with its scope. When an object for the Java bean does not exist within the declared scope, the bean is instantiated. Java beans can be declared with the scope of page, request,session or application. ● jsp:getProperty: Calls the get method of a Java bean object and retrieves the property value. ● jsp:setProperty: Calls the set method of a Java bean object and sets the property value. Java beans must be declared with the jsp:useBean tag before the other two operations. This means that a declaration of a Java bean and all of the other operations for the Java bean must be included in the same JSP page. This strongly limits the extent of a JSP fragment. In order to relax the limitation, JSP splitting allows declarations of Java beans to be copied to JSP fragments. The reason why declarations of Java beans can be copied is that the declarations do not have side effects. Each declaration merely instantiates a Java bean object when the Java bean object has not yet been instantiated within its scope. In other words, a Java bean object is not instantiated more than one time within its scope even if the declaration is copied to JSP fragments. Therefore, some of the operations for a Java bean can be split into fragments. When a property value is set for a Java bean by using a jsp:setProperty tag and retrieved from the JSP bean by using a jsp:getProperty tag, a true data dependence exists between the two the rule that data dependent statements cannot be split is applied, data dependent operations for a Java bean must be included in the same fragment. However, a data dependence between an operation that sets a property value for a JSP bean and another operation that retrieves the property value is maintained after the two operations are split if the following two conditions are satisfied. Cond. 2 1 The scope of the Java bean is ”request”, ”session”, or ”application”. Cond. 2 2 The split JSP pages are executed in the same JSP engine. If these conditions are satisfied, JSP splitting allows the operations among which the data dependences exist to be split into different JSP pages, because the Java bean lives in the JSP engine during the execution of the split JSP pages. As a result, the portions split along with the portions selected by a developer can be reduced. However, the Web page fragments obtained by executing the split JSP pages satisfying these conditions must have the same caching period. This is because the data dependence is not maintained if a JSP page was executed and a value is defined in the JSP page while another JSP page, in which the defined value is used, was not executed due to the valid cached content of the Web page. When a Java bean has the page scope, the dependent operations for the Java bean cannot be split. However, each operation that sets a property value for a Java bean with the page scope can be copied along with one or more operations that retrieve the same property value, because the operations that set property values for Java beans with the page scope do not have side effects. Figures 3 to 6 show the algorithms to split the operations for Java beans. These algorithms try to minimize the number of the operations that are moved or copied to a JSP fragment. 4. JSP Splitting Tool We implemented JSP splitting as a plugin for Eclipse。 我們使用 GUI 工具實(shí)施 JSP 切分 ，并且證實(shí) 切分的JSP 片段和 JSP 頁(yè)面 切分 之前是以同一方式執(zhí)行的。由 Apache Jakarta 項(xiàng)目提供的 存儲(chǔ) 標(biāo)簽庫(kù)也為 JSP 片段支持不同的存儲(chǔ)方法。例如框架， 橫幅 ， 標(biāo)題等 。另外，由于 數(shù)據(jù)在 JSP 頁(yè)面 的執(zhí)行 中被傳送到 JSP 引擎， JSP 引擎的源程序，例如 Apache Tomcat，為了正確的分析數(shù)據(jù)依賴(lài) 是必須的 。章節(jié) 6 討論有關(guān)我們的貢獻(xiàn)的 先 前的工作。應(yīng)用卸載技術(shù)允許 主 JSP 頁(yè)面和JSP 片段 在 不同的服務(wù)器上執(zhí)行。 條件 . 11 主 JSP 頁(yè)面和 JSP 片段的最終執(zhí)行結(jié)果要和 JSP 頁(yè)面被 切分 之前一致。因?yàn)?JSP 頁(yè)面 通過(guò)使用如 Apache Jasper 編譯器那樣的編譯器被編譯成 servlet 程序 ， JSP 切分 能夠通過(guò)應(yīng)用程序分化技術(shù)到 JSP 頁(yè)面中編譯的 servlet中 獲得。因此，程序 切片 技術(shù)對(duì)于 切分 JSP 頁(yè)面沒(méi)有做的很好。然而，如果這個(gè)規(guī)則太嚴(yán)格的被應(yīng)用，在圖 1（ b）中顯示的申明沒(méi)有能被分化的，因?yàn)閺?S1 到 S9 開(kāi)始有交錯(cuò)的依 賴(lài)。這意味著 out的值能夠被一些 JSP頁(yè)面 使 用并且 寫(xiě)入這個(gè)值的文本數(shù)據(jù)由 JSP引擎 維持 。 一般來(lái)講，無(wú)序執(zhí)行對(duì)于沒(méi)有數(shù)據(jù)控 制依賴(lài)的申明是允許的。 Java beans能夠申明在頁(yè)面，請(qǐng)求，節(jié)或者應(yīng)用的范圍內(nèi)。為什么 Java beans的申明能夠被復(fù)制 的原因是 申明沒(méi)有副作用。 條件 . 22 切分 的 JSP頁(yè)面在同一 JSP引擎中執(zhí)行。圖 7 顯示了 JSP 切分 工具的結(jié)構(gòu)。這時(shí)，在 JSP 頁(yè)面中的每一部分和 servlet 程序中的一個(gè)或多個(gè)申明之間的映射表被形成。 另外， 我們是在網(wǎng)頁(yè)沒(méi)有存儲(chǔ)的時(shí)候測(cè)試執(zhí)行性能的。 Extra Method 是一種軟件重構(gòu)技術(shù)， 它包含了若干程序聲明在一個(gè)新的方法中