D. Batory and B. J. Geraci, "Composition validation and subjectivity in GenVoca generators," IEEE Transactions on Software Engineering (special issue on Software Reuse), pp. 62--87, 1997.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....compilation. As a result, run time adaptability is not supported. Other feasible approaches to handle dynamic weaving is to implement aspects using meta objects. DJ Aspects [21] is a proposal that addresses dynamic weaving through the use of metaprogramming. Other approaches include the AMF [2, 4, 5] that provides dynamic weaving in a framework. We believe that ideally an implementation should support both static and dynamic weaving. 3.2 Level of weaving and life span of aspects The level of weaving defines the point up to which one manages (or wishes) to achieve separation of concerns in ....

.... by Contract (DBC) 19] Under this principle, a software system is viewed as a set of communicating components whose interaction is based on precisely defined specifications of some mutual obligations (contracts) The concept of deploying assertions and DBC in the context of AOP was discussed in [2, 3, 13]. DBC was introduced in the context of the Eiffel programming language, where a contract is hard wired in the class definition and it is inherited. Further, in iContract [14] assertions are propagated along inheritance and interface relationships. Separating contracts from the functional ....

[Article contains additional citation context not shown here]

D. Batory and B.J. Geraci. Composition Validation and Subjectivity in GenVoca Generators. In IEEE Transactions on Software Engineering, pages 67 -- 82, 1997.

....the product line. The issues involved in creating a development process and business environment tailored to the use of a product line architecture are relatively well understood [4] Additionally, representations for specifying and storing product line architectures have already been developed [3,10,14,34]. Effective use of a particular product line architecture, however, also requires a support environment to manage its evolving structure an area of research that has largely been ignored to date. This paper introduces Mnage, an environment that is specifically designed to fill this void. Mnage ....

....include C2SADEL [20] Darwin [19] Rapide [17] UniCon [27] and Wright [1] A number of these languages also provide extensive support for modeling behaviors and constraints on the properties of components and connectors [21] However, with the exception of xADL 2. 0 [10] Koala [34] GenVoca [3] and to some extent Acme [14] existing architecture description languages do not directly support the specification of product line architectures. Given the importance of prod uct lines in today s world of software development, we expect this situation to change rapidly. 3. Problem It is well ....

D. Batory and B.J. Geraci, Composition Validation and Subjectivity in GenVoca Generators. IEEE Transactions on Software Engineering, 1997. 23(2): p. 67-82.

....various approaches exist (e.g. GenVoca [11] Koala [112] there is a need for more general, languageindependent solutions. Moreover, configuration validation, for instance by modeling configuration constraints, is needed to automatically detect and prevent invalid component configurations [51, 8, 9]. Abstractions for component compositions can be domain specific and are either technical or consumer related. The group of products (or systems) that can be built from technical abstractions forms a product family [115] or system family) The group of products that can be built from ....

D. Batory and B. J. Geraci. Composition Validation and Subjectivity in GenVoca Generators. IEEE Transactions on Software Engineering (special issue on Software Reuse), pages 62--87, 1997. 151

....engines and type systems. We expect further work with a richer set of designs to help us identify when the full power of our current methodology is required. Feature oriented designs can benefit from a broader scope of verification techniques. Early work on analyzing dependencies between features [3] must be formalized and incorporated into any validation framework. Addressing this problem may involve creating special architectural description languages for feature oriented designs that capture these dependencies and simplify construction of the exit and reentry subgraphs. Finally, we have ....

D. Batory and B. J. Geraci. Composition validation and subjectivity in GenVoca generators. IEEE Transactions on Software Engineering, pages 67--82, Feb. 1997.

....has become particularly clear in the area of software product lines, where the goal is to construct large families of related applications from primitive and reusable components. The components that made this possible encapsulated fragments of classes. We use the term refinement (also in [BG98] for any such unit of functionality in a software system. A refinement is a functionality addition to a program that introduces a conceptually new service, capability, or feature, and may affect multiple implementation entities. Various researchers have offered different descriptions, ....

....[Hol92] for this example are, in fact, general enough to be applicable to a directed graph. If, however, a more efficient, specialized for undirected graphs algorithm was used (as is, for instance, possible for the Cycle Checking operation) the change would yield incorrect results. Sma99, SB98b, BG98] discuss in detail the issue of ensuring that collaborations are actually interchangeable. Although we have focussed on a single application that supports all three graph operations, it is easy to see how variants of this application could be created (e.g. by omitting some operations or adding ....

[Article contains additional citation context not shown here]

D. Batory and B.J. Geraci. Composition Validation and Subjectivity in GenVoca Generators, IEEE Transactions on Software Engineering, February 1997, 67-82.

....other works which consider flexible composition schemes, but only in the context of object oriented programming languages [Prehofer01, Elrad01] but do not consider graphic modeling of components. Our composition methods whit layers resembles other layered architectures for feature composition [Batory97, Demeter96], which lack explicit support for interactions. 5. ....

D. Batory and B. J. Gerac. Composition validation and subjectivity in Genvoca generators. IEEE Transactions on Software Engineering, February 1997.

.... 1990] 2NIX [Barstow, 1986b, Barstow, 1991] and PSI system [Barstow, 1986a] Jomponent Based Methods: System implemented based on this approach rely on libraries of parametrized, plug compatible, and reusable components to transform high level specifications of target systems into source code [Batory and Geraci, 1997]. Ap plication generators based on a library of patterns, like the Pattern Based Simulator Generator (PSiGene) Schuetze et al. 1997] are typical examples. Rich and Waters refer to these approaches as Itspectiot Methods [Rich and Waters, 1988] According to them, human programmers seldom think ....

D. Batory and B.J. Geraci. Composition validation and subjectivity in GenVoca generators. IEEE Transactions on Software Engineering, 23(2):67 82, February 1997.

....but that is of little consequence) and because the correctness of the translation from algorithm to code is at least as reliable as compiler technology, the complexity of the task at hand is greatly ameliorated. Namely, components are expected to live up to certain contractual obligations [1, 4, 11]. In the case of a library constructed entirely through the methodology presented here, these components would be composed in like manner so as to make this task manageable. This is largely due to the fact that the approach presented here leads to a software architecture layered in such a way so ....

Don Batory and Bart J. Geraci. Composition validation and subjectivity in genvoca generators. IEEE Transactions on Software Engineering (special issue on Software Reuse, pages 67-82, 1997.

....realisation for separated concerns. In contrast to existing product line engineering methods the bottom up approach is oriented on the reuse and extension of existing systems. However there had been approaches to apply aspect oriented principles to build product lines. One approach is GenVoca [2] which allows to build a system family as layered architecture. Another approach may be found in [4] This paper proposes the application of aspects in product line software development in general. Complementary to the bottom up approach is the field of feature interaction which provides means ....

D. Batory and B.J. Geraci. Composition Validation and Subjectivity in GenVoca Generators. In IEEE Transactions on Software Engineering, pages 67 -- 82, 1997.

....subjective way. A complete software system is then composed out of subjects by writing the composition rules, which specify the correspondence of the subjects (i.e. namespaces) classes and members to be composed and how to combine them. As Czarnecki [7] observes, this is close to GenVoca approach [2, 3], where the systems are composed out of layers according to the design rules (for further information on this topic see [12] GenVoca layers can be easily simulated as subjects. SOP can be viewed as a special case of AOP where the aspects according to which the system is being decomposed are ....

Don Batory and Bart J. Geraci. Composition validation and subjectivity in GenVoca generators. IEEE Transactions on Software Engineering (special issue on Software Reuse), pages 67-82, February 1997. See http://www.cs.utexas.edu/users/schwartz/pub.htm.

....of issues discussed in this paper will need to be addressed for all components, regardless of whether they are generic. The performance specification problem takes on an additional dimension of complexity when reusing a parameterized data abstraction whose implementation has not been selected yet [1, 17], because the performance of the new component needs to be parameterized to allow any implementation of the reused data abstraction to be chosen. Given the potential complexity of parameterized performance expressions, it may be necessary to define one or more hypothetical situations and then ....

Batory, B., and Geraci, B. J., "Composition Validation and Subjectivity in GenVoca Generators", IEEE Transactions on Software Engineering, 23(2): 67-82, February, 1997.

....a product line, where each equation corresponds to a member of that product line. Since not all features are compatible [7] there are constraints on how units of a model can be legally composed. These constraints, called design rules, are not essential to this paper; details are described in [5]. Implementation. Figure 1a b show specifications K and R. K defines a base class C; R defines a class refinement which adds variable y and method h to class C. In general, a class refinement can add new data members, methods, constructors, as well as extend existing methods and constructors. The ....

....to build suites of Jak and Bali tools and envision additional matrices to synthesize tools for other AHEAD program representations. An example is design rule files. A design rule file specifies the preconditions and postconditions for unit usage; it is written in a special predicate language [5]. We now have a tool, called drc, which composes design rule files and reports composi base b2jak b2jcc bc core parser codegen cgen cgen require reqGram reqb2jak reqb2jcc reqComp tool b2jktool b2jcctool composr Table 5. Bali Matrix tion errors. We anticipate that additional tools for design ....

D. Batory and B.J. Geraci, "Composition Validation and Subjectivity in GenVoca Generators", IEEE Transactions on Software Engineering, Feb. 1997, 67-82.

....it is too expensive to build all possible family members; it is much cheaper to build components and to assemble desired family members from them. Many methodologies have been invented to create product line architectures and several technologies have been used to implement them (e.g. [2, 4, 10, 12]) In this paper, we explore how AOP techniques could fit in the product line context. To do that we selected AspectJ [1, 7, 8] and applied it to the problem domain of graph algorithms [9] which is relies on common knowledge yet it is complex enough to highlight the key points of a product line ....

....addEdge (e) Figure 5. Adding weighted edges in Weighted aspect 5 Findings In product line designs it is the case that not all syntactically valid composition of features are semantically valid. The legal compositions of features in Table 1 are defined by simple constraints called design rules [2]. In AspectJ there is no support for design rules, that is, the prograrmner has to manually select all the files necessary to create a new member, this activity is complex and error prone even for small product lines. 7 Aspects can only introduce classes that are private to the aspects and that ....

D. Batory and B. Geraci. Composition Validation and Subjectivity in GenVoca Generators. IEEE Transactions on Software Engineering, February 1997.

....typing constraints do exist. Design rules capture syntactic and semantic constraints that govern legal compositions. It is common that the selection of a feature will disable or enable the selection of other features. The details of design rules are not germane to our paper and can be found in [3]. 2.2 Model Implementation A GenVoca constant is a set of classes. Figure 2 depicts a constant i that encapsulates four classes (adi) A Gen Voca function is a set of classes and class refinements. A class refinement adds new data members, methods, and extends or overrides existing methods of a ....

D. Batory and B.J. Geraci, "Composition Validation and Subjectivity in GenVoca Generators", 1EEE Transactions on Software Engineering, Feb. 1997, 67-82.

....hypermodule without any trouble, since both search hyperslices define and use the same data members and methods. However, evidently, the outcome of the execution of the algorithm will be incorrect. The legal compositions of features in Table 1 are defined by simple constraints called design rules [1]. In Hyper J there is no support for design rules, that is, the programmer has to manually select all the files necessary to create a new member and include them in the correct order in the hypermodule, this activity is complex and error prone even for small product lines like GPL. Declarative ....

D. Batory and B. Geraci. Composition Validation and Subjectivity in GenVoca Generators. IEEE Transactions on Software Engineering, February 1997.

....behavioral properties. Thus, it is common that not all combinations of features (or their implementations) are correct [25] A model for expressing program types and algorithms that can automatically and efficiently validate equations have been developed and are part of the Jakarta Tool Suite [5]. 2.2.3 Mixin Layer Implementation There are many ways in which to implement refinements, ranging from dynamically composing objects to statically composed meta programs (i.e. programs that generate other programs)t4] and rule sets of program transformation systems [31] One of the simplest is ....

D. Batory and B.J. Geraci, "Composition Validation and Subjectivity in GenVoca Generators", 1EEE Transactions on Software Engineering, February 1997, 67-82.

....for productline architectures is one of economics and practicality: it is too expensive to build all possible family members; it is much cheaper to build components and to assemble desired family members from them. Many methodologies have been invented to create product line architectures (e.g. [Ame00, Bat97, Bos99, Coh98, Cza99, Cza00, DeB99, Gom94, Par76, Wei99]) Unfortunately, the state of the art is immature. We are unaware of any attempts to evaluate different methodologies on a common set of problems. If this were done, we would understand better the strengths and weaknesses of different methodologies. We would know when to use a particular ....

....functions for profiling. Prog creates the objects required to represent a graph, and calls the algorithms of the family member on this graph. Refinements can not be composed in arbitrary orders. The legal compositions of refinements in Table 2 is defined by simple constraints called design rules [Bat97] whose details we omit from this paper, but do include with our source code. Our GUI specification tool translates a sentence in the grammar of Figure 1 (in addition to checking for illegal combinations of features) into an equation. Because features are in 1 to1 correspondence with refinements, ....

D. Batory and B. Geraci. Composition Validation and Subjectivity in GenVoca Generators. IEEE Transactions on Software Engineering, February 1997.

No context found.

D. Batory and B. J. Geraci, "Composition validation and subjectivity in GenVoca generators," IEEE Transactions on Software Engineering (special issue on Software Reuse), pp. 62--87, 1997.

No context found.

Batory, D., Composition Validation and Subjectivity in GenVoca Generators, Software Engineering, (1997).

No context found.

D. Batory, B. J. Geraci, "Composition Validation and Subjectivity in GenVoca Generators", IEEE Trans. Softw. Eng., Vol. 23, No. 2, pp. 67-82, 1997.

No context found.

D. Batory and B. J. Geraci. Composition validation and subjectivity in GenVoca generators. IEEE Transactions on Software Engineering, 23(2):67--82, Feb. 1997.

No context found.

D. Batory and B. J. Geraci. Composition validation and subjectivity in GenVoca generators. IEEE Transactions on Software Engineering, 23(2):67--82, Feb. 1997.

No context found.

D. Batory and B. J. Geraci, "Composition Validation and Subjectivity in GenVoca Generators," IEEE Transactions on Software Engineering, pp. 67-82, Feb. 1997.

No context found.

D. Batory and B. J. Geraci, "Composition validation and subjectivity in genvoca generators," in IEEE Transactions on Software Engineering, pp. 67--82, feb 1997.

No context found.

D. Batory and B.J. Geraci. Composition Validation and Subjectivity in GenVoca Genera5 tors. In IEEE Transactions on Software Engineering, pages 67 -- 82, 1997.

First 50 documents

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