D. W. Binkley. Computing amorphous program slices using dependence graphs and a data- model. In ACM Symposium on Applied Computing, pages 519-525, The Menger, San Antonio, Texas, U.S.A., 1999. ACM Press, New York, NY, USA.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....program s semantics [2, 5, 9, 12, 17] Traditionally, this projection is de ned in terms of a subset of variables of interest and is constructed using the sole transformation of statement deletion [17] The slice is therefore a subprogram which preserves a subcomputation. Amorphous slicing [1, 8, 10] is a variation of traditional slicing, in which the semantic properties are retained (the slice maintains a subcomputation) while the syntactic restriction to statement deletion is relaxed. By sacri cing the connection with syntax, smaller slices can be produced. Amorphous slicing is therefore ....

....which produce good quality (i.e. small) slices in reasonable time. The construction of amorphous slices is harder than the construction of syntax preserving slices, because any transformation can potentially be applied in amorhous slicing. Previously published algorithms for amorphous slicing [1, 10] have concerned only intraprocedural languages. In this paper we present a simple interprocedural amorphous slicing algorithm and initial results on the performance of the dependence reduction transformation at the heart of our amorphous slicing algorithm. Our current approach simply pushes the ....

David Wendle Binkley. Computing amorphous program slices using dependence graphs and a data- ow model. In ACM Symposium on Applied Computing, pages 519-525, The Menger, San Antonio, Texas, U.S.A., 1999. ACM Press, New York, NY, USA.

.... variables [6, 14, 20, 36, 44] There are various forms of slicing including the original static formulation [43] and subsequent dynamic [1, 27] quasi static [38] conditioned and constrained [7, 10, 18] and pre post conditioned [21] Slices can be construed to be syntax preserving or amorphous [5, 19, 40]. The problem of computing static variable dependence can be partly solved by a solution to the corresponding static slicing question. This result can be used in order to determine control flow. That is, in order to decide whether p is in the control dependence relation for the variable x in if ....

BINKLEY, D. W. Computing amorphous program slices using dependence graphs and a data-flow model. In ACM Symposium on Applied Computing (The Menger, San Antonio, Texas, U.S.A., 1999), ACM Press, New York, NY, USA, pp. 519--525.

....Further work is required to extend the expressibility of the slicing criterion and to provide partially automated assistance to criterion selection. 2. Domain Speci c Transformation component Initial work on domain speci c transformation rule sets considered problems in robustness veri cation [5, 13] and dynamic memory modelling [16] Further work is required to develop additional targeted transformation rules sets. 3. Empirical evaluation of amorphous slicing The theory and applications of amorphous slicing to comprehension, reuse, testing and debug8 Rule 1 (assignment) e] s) v s ....

David W. Binkley. Computing amorphous program slices using dependence graphs and a data- ow model. In ACM Symposium on Applied Computing, pages 519-525, The Menger, San Antonio, Texas, U.S.A., 1999. ACM Press, New York, NY, USA.

....but contingent upon the values of its variables, Isabelle can sometimes derive a simplification of the condition. Future versions of ConSIT will exploit this, substituting the simplified condition when conditioning the program. The intention of this work is to develop amorphous slicing capability [16, 3]. 5. Examples 5.1. The Example from Canfora, Cimitile and DeLucia Consider again, the example program in Figure 1. Figure 3 shows the conditioned slice produced by the ConSIT system. This is identical to the conditioned slice produced by Canfora, Cimitile and DeLucia using a mixture of automated ....

....is somewhat limited. Pointers and procedures are the most notable omissions. Work is in progress to extend ConSIT to handle fully interprocedural conditioned slicing. Another problem for future work is the adaption of the technologies used to implement ConSIT so that amorphous conditioned slices [16, 18, 3] can be computed. Amorphous slices share the semantic restriction that they preserve the effect of the subject program on the slicing criterion, but are syntactically unrestricted. In cases where the Isabelle theorem prover was unable to decide the truth or otherwise of propositions put to it, it ....

D. W. Binkley. Computing amorphous program slices using dependence graphs and a data-flow model. In ACM Symposium on Applied Computing, pages 519--525, The Menger, San Antonio, Texas, U.S.A., 1999. ACM Press, New York, NY, USA.

....because the mutants socreated are guaranteed to be equivalent. This is slightly inaccurate [12] The observation becomes true when only reference preserving mutations are considered. The present authors [12] developed the initial proposal of Voas and McGraw, showing how amorphous slicing [3, 8, 9, 11] can be used to assist human analysis of particularly stubborn mutants. This work considered slicing as a means of assisting the human analysis, rather than as a way of determining equivalence. Offut et al. 19, 20, 21, 22, 23] have considered the problems of test data generation and equivalent ....

BINKLEY, D. W. Computing amorphous program slices using dependence graphs and a data-flow model. In ACM Symposium on Applied Computing (The Menger, San Antonio, Texas, U.S.A., 1999), ACM Press, New York, NY, USA, pp. 519--525.

....an approximate answer. In the next section the way in which the constraint based and slicing based approach may complement is described. Currently, implementations of amorphous slicing are at the level of proof of concept systems, such as those described in ( Harman and Danicic, 1995] [Binkley, 1999]) and ( Harman et al. 1998] These algorithms produce good amorphous slices in specialised domains. However, their quadratic complexity means that they cannot be expected to scale up to the problem of equivalent mutants for large programs 4 . However, more recently work has begun on an ....

.... mutation equivalence (inserting the assignments to z) giving z = TRUE; x=y; s=x 1; k=k 2; if(s 0) z = z y 2 = x 2; k=y 2; Conventional (non amorphous) end slicing on fzg gives: z = TRUE; x=y; s=x 1; if(s 0) z = z y 2 = x 2; Constant propagation using Binkley s algorithm [Binkley, 1999] gives z = TRUE; if(s 1 0) z = TRUE y 2 = y 2; At this point boolean simpli cation is required [Harman et al. 1999, Harman et al. 1998] This boolean simpli cation could be viewed as trivial theorem proving demonstrating the truth of a simple theorem. The theorem in question is ....

Binkley, D. W. (1999). Computing amorphous program slices using dependence graphs and a data- ow model. In ACM Symposium on Applied Computing, The Menger, San Antonio, Texas, U.S.A. ACM Press, New York, NY, USA. to appear.

....which avoids the code growth, but is limited by the program s original calling structure, in particular, making conservative assumptions at the convergence of paths in the call graph. 2.3. 1 Interprocedural Data Flow Analysis There exists much research on interprocedural data flow analysis [9, 11, 13, 18, 30, 33, 37, 43, 44, 63] for aliasing, side effect analysis, constant propagation, program slicing, and other uses. These techniques involve performing an analysis of the whole program code, gathering information about conditions across procedural boundaries. For example, variable definitions and uses may be tracked as ....

....making the region appear to the rest of the compiler as a function. After compilation and optimization, a region is reintegrated into the containing function by updating changes in data flow conditions (i.e. variable 2 Another interesting and analogous technique called amorphous program slicing [9] has some similar goals to region based compilation, including the extraction and amalgamation of related but potentially scattered portions of program code and the control of code growth. Although intended for higher level software engineering tasks, the ideas it introduces may be valuable to my ....

D. Binkley. Computing amorphous program slices using dependence graphs. In Proceedings of the 1999 Symposium on Applied Computing, pages 519--525, San Antonio, Texas, Feb 28-Mar 2, 1999.

....but contingent upon the values of its variables, Isabelle can sometimes derive a simplification of the condition. Future versions of ConSIT may exploit this, substituting the simplified condition when conditioning the program. The intention of this work is to develop amorphous slicing capability [4, 17]. 5. Examples 5.1. The Example from Canfora, Cimitile and DeLucia Consider again, the example program in Figure 1. Figure 3 shows the conditioned slice produced by the ConSIT system. This is identical to the conditioned slice produced by Canfora, Cimitile and DeLucia using a mixture of automated ....

....is somewhat limited. Pointers and procedures are the most notable omissions. Work is in progress to extend ConSIT to handle fully interprocedural conditioned slicing. Another problem for future work is the adoption of the technologies used to implement ConSIT so that amorphous conditioned slices [4, 17, 19] can be computed. Amorphous slices share the semantic restriction that they preserve the effect of the subject program on the slicing criterion, but are syntactically unrestricted. In cases where the Isabelle theorem prover was unable to decide the truth or otherwise of propositions put to it, it ....

D. W. Binkley. Computing amorphous program slices using dependence graphs and a data-flow model. In ACM Symposium on Applied Computing, pages 519--525, The Menger, San Antonio, Texas, U.S.A., 1999. ACM Press, New York, NY, USA.

....value, as mutation testing is a unit level testing technique and is therefore likely to be applied to small code units, requiring, in many cases, only small scale intraprocedural slicing. More recently work has begun on an implementation of amorphous slicing using the Program Dependence Graph ([Binkley, 1999]) The aim of this work is to produce realistic, ecient, scalable algorithms for amorphous slicing. 9 A Comparison of Constraint and Amorphous Slicing Based Approaches This section explores the relationship between the two approaches of amorphous slicing and constraint solving in detection of ....

.... equivalence (inserting the assignments to z) giving z = TRUE; x=y; s=x 1; k=k 2; if(s 0) z = z y 2 = x 2; k=y 2; Conventional (non amorphous) end slicing on fzg gives: 17 z = TRUE; x=y; s=x 1; if(s 0) z = z y 2 = x 2; Expression propagation using Binkley s algorithm [Binkley, 1999] gives z = TRUE; if(y 1 0) z = TRUE y 2 = y 2; At this point boolean simpli cation is required [Harman et al. 1999, Harman et al. 1998] This boolean simpli cation could be viewed as trivial theorem proving demonstrating the truth of a simple theorem. The theorem in question is ....

Binkley, D. W. (1999). Computing amorphous program slices using dependence graphs and a data- ow model. In ACM Symposium on Applied Computing, pages 519-525, The Menger, San Antonio, Texas, U.S.A. ACM Press, New York, NY, USA.

....value of a pseudo variable. This was an early form of what is termed safety slicing in the present paper. It used a combination of slicing and ah hoc transformations, but the term amorphous slice was not coined until 1997 [Harman and Danicic, 1997] and detailed algorithms did not appear until 1999 [Binkley, 1999]. In 1997, Gerber and Hong [Gerber and Hong, 1997] described an approach to scheduling which relies upon slicing to identify the observable parts of a real time system (those that interact with the environment through I O operations) and the unobservable components (those which do not perform ....

....by National Science Foundation grant CCR 9803665. Sebastian Danicic, in part, is supported by Engineering and Physical Sciences Research Council grant GR R98938 and GR M58719. Preliminary versions of some of the material presented here have appeared in conferences [Harman and Danicic, 1997,Binkley, 1999,Binkley et al. 2000] 7 Summary In traditional approaches to slicing, the only simplifying transformation used to create slices is component deletion. This choice was motivated by the original application of slicing to debugging, where the syntax preserving nature of a slice was important. ....

Binkley, D. W. (1999). Computing amorphous program slices using dependence graphs and a data- ow model. In ACM Symposium on Applied Computing, pages 519-525, The Menger, San Antonio, Texas, U.S.A. ACM Press, New York, NY, USA.

....values are of interest. More recent work has extended traditional slicing by considering novel slicing criteria involving conditions and test adequacy properties [5, 15] The techniques for isolation of statements have also broadened from statement deletion to allow for more general transformation [4, 12, 30]. This paper will be concerned solely with syntaxpreserving static slicing, which will be used both to re ne and to extend the results of concept assignment. In all cases, slices will be constructed for a set of nodes of a program s Control Flow Graph (CFG) This means that the slicing criterion ....

D. W. Binkley. Computing amorphous program slices using dependence graphs and a data- ow model. In ACM Symposium on Applied Computing, pages 519-525, The Menger, San Antonio, Texas, U.S.A., 1999. ACM Press, New York, NY, USA.

....tests whether or not the hypothesis is valid, not for some speci c input, but statically at compile time, and therefore, in general. The test harness is constructed from pre and post conditions and the resulting composition of test harness and subject program are simpli ed using amorphous slicing [15, 16, 6]. The formal component consists, not of attempting to prove the proposition, but in attempting to simplify the composition of the test harness and the subject program. This combination creates a method for approximating the answers to undecidable propositions using program simpli cation ....

....of Steps 3.1 and 3.2. In [16] Step 3.1 is implemented using a simple symbolic executor (implemented using assignment push transformations) directly on the program s Abstract Syntax Tree, and Step 3. 3 is implemented using an implementation of the parallel slicing algorithm described in [11] In [6] both steps are applied directly to the Program Dependence Graph (PDG) 17, 13] Symbolic execution is replaced by a graph unfolding process and a data ow interpretation of the PDG. The domain speci c transformations of Step 3.3 are written in terms of the PDG also. This has the advantage that ....

[Article contains additional citation context not shown here]

Binkley, D. W. Computing amorphous program slices using dependence graphs. In ACM Symposium on Applied Computing (The Menger, San Antonio, Texas, U.S.A., 1999), ACM Press, New York, NY, USA. to appear.

No context found.

D. W. Binkley. Computing amorphous program slices using dependence graphs and a data- model. In ACM Symposium on Applied Computing, pages 519-525, The Menger, San Antonio, Texas, U.S.A., 1999. ACM Press, New York, NY, USA.

No context found.

D.W. Binkley, "Computing Amorphous Program Slices Using Dependence Graphs and a Data-Flow Model," Proc. ACM Symp. Applied Computing, pp. 519-525, 1999.

No context found.

D. W. Binkley. Computing amorphous program slices using dependence graphs and a data-flow model. In ACM Symposium on Applied Computing, pages 519--525, The Menger, San Antonio, Texas, U.S.A., 1999. ACM Press, New York, NY, USA.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC