M. Lanza, "Combining Metrics and Graphs for Object Oriented Reverse Engineering," diploma thesis, Univ. of Bern Oct. 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....about the potential consequences of making a change. These tools are typically proactive, because they intend to calculate the effect of a change beforehand. Tools that detect the need for architectural evolution find out when, where and how the architecture should evolve or be restructured [8,16]. We also need tools that provide disciplined support for architectural evolution, and try to detect and resolve conflicts during evolution [4,17] The need for co evolution arises when the architecture, design and implementation can evolve independently, but need to be kept synchronized in ....

M. Lanza. Combining Metrics and Graphs for Object Oriented Reverse Engineering. Thesis, University of Bern, October 1999.

....The actual target can only be resolved at run time. 3 Embedding Implementation Concepts into UML Given the core object oriented implementation model depicted in Figure 2, the question is whether this can be embedded into UML. Comparing the core implementation model with the UML meta model [17] [18] we make the following observations (summarised in Figure 3) a) The UML meta model defines a large number of concepts that do not appear in the implementation model. Aggregation and Constraint are two examples but there are many more. b) There is a substantial overlap between the core ....

....of a method invocation and an attribute access in UML. In the following subsections we will analyse some possibilities to extend the UML meta model to incorporate these two concepts. 3. 1 The Behavioural Elements Approach There are a number of UML concepts in the Behavioural Elements package (see [18], part 3) that come close to the Invocation and Access concepts of the implementation model (see Figure 2) It is out of the scope of this paper to describe all possibilities in detail, so we restrict ourselves to an analysis of the most plausible concept, namely Action . This concept is ....

[Article contains additional citation context not shown here]

M. Lanza. Combining metrics and graphs for object oriented reverse engineering. Diploma thesis, University of Bern, Oct. 1999.

No context found.

M. Lanza, "Combining Metrics and Graphs for Object Oriented Reverse Engineering," diploma thesis, Univ. of Bern Oct. 1999.

No context found.

M. Lanza, "Combining metrics and graphs for object oriented reverse engineering," Diploma thesis, University of Bern, October 1999.

No context found.

M. Lanza. Combining Metrics and Graphs for Object Oriented Reverse Engineering. Diploma thesis, University of Bern, October 1999. (pp 20, 28, 29, 49, 50, 67, 77)

....tool called CodeCrawler, whose development started in 1998 and which has been re implemented several times since then and is currently still being developed and extended. CodeCrawler is a lightweight software visualization tool which combines metrics information with its visualizations [23, 9, 24]. It is mainly targeted at static software visualization. CodeCrawler s implementation changed and evolved in order to cope with the new requirements our research generated. Its current design is thus able to solve most of the problems we encountered with static software visualizations. Moreover, ....

....components can be regarded as the overall architecture of a software visualization tool. 3 CodeCrawler CodeCrawler is a language independent software visualization tool written in Smalltalk. CodeCrawler supports reverse engineering through the combination of metrics and software visualization [23, 9, 12, 24]. In Figure 1 we can see a screen shot of CodeCrawler. Its power and flexibility, based on simplicity and scalability, has been repeatedly proven in several large scale industrial case studies. To model software, CodeCrawler uses Moose, a language independent reengineering environment. The first ....

[Article contains additional citation context not shown here]

M. Lanza. Combining metrics and graphs for object oriented reverse engineering. Diploma thesis, University of Bern, Oct. 1999.

....more detail in the next section. The Moose Reengineering Environment included a metrics engine, which was first implemented to perform simple queries and then greatly extended to support the development of CodeCrawler, a software visualization tool which integrates metrics in its visualizations [18, 8]. The extended metrics engine included a suite of language independent metrics, whose computation was based exclusively on the language independent metamodel representation of software artifacts. As we are now starting the implementation of a metametamodel, where apart from artifacts defined in ....

....we allow for language plugins that extend the model with language specific features. Next to that, we allow tool plug ins to extend as well the model with tool specific information. 4. 2 CodeCrawler CodeCrawler supports reverse engineering through the combination of metrics and visualization [18, 8, 19]. It visualizes entities as nodes and relatioships as edges, and renders up to five metrics on the nodes by using their width, height, color and position on the display. Through these simple visualization enriched with the large metrics suite provided by Moose, it enables the user to gain insights ....

Michele Lanza. Combining metrics and graphs for object oriented reverse engineering. Diploma thesis, University of Bern, October 1999.

....can also make use of it and a bigger ease of expression. # A Metrics Framework. Most of the queries we perform are based on metric properties of the entities. For that purpose we have implemented a large framework of metrics (at this time more than 50) which is better explained in [9]. 3 A Taxonomy of Queries In this section we explain what kinds of queries we can build and how they can be composed into more complex ones. Note that the notation we use in this paper does not reflect the actual notation we use, which is much more verbose. For the sake of simplicity and ....

....of a paper with the major results obtained with this approach applied on several large and very large case studies. Furthermore we will extend the query engine and its query language to render it as flexible and powerful as possible. We also plan to use the software visualization tool CodeCrawler [9, 2, 5] in this context. ....

M. Lanza. Combining metrics and graphs for object oriented reverse engineering. Diploma thesis, University of Bern, Oct. 1999.

....3.3 Scalability issues Calculating simple differences between the versions of a system produces a huge amount of data. The treatment and condensing of this information into a meaningful content is crucial for providing useful information. In the first experiment we made, we use CODE CRAWLER [3, 8] to help us to visualize the results of the queries. 4 Case Study HotDraw[1] is a two dimensional graphics framework for structured drawing editors, implemented in Smalltalk by John Brant. You can easily create new graphical objects and special manipulation tools for your drawings. HotDraw also ....

M. Lanza. Combining metrics and graphs for object oriented reverse engineering. Master's thesis, University of Bern, 1999.

....Classes using Metrics As we have previously seen we use two dimensional boxes to represent classes. We can enrich this representation using metrics: the width and height of the boxes reflect metric measurements of the classes, as we see in Figure 3. This approach has been presented in [LAN 99] and [DEM 99a] In the visualization presented in this paper we visualize classes and use the metrics number of methods (NOM) for the width and number of attributes (NOA) for the height, although in our tool we can choose other metrics. In the case of the differential metrics we just subtract ....

....same system using the differential view which emphasizes the changes made. 5. Implementation: CodeCrawler and Moose CodeCrawler is the tool used to generate the views presented in this paper. CodeCrawler supports reverse engineering through the combination of metrics and software visualization [LAN 99, DEM 99a, DUC 01] Its power and flexibility, based on simplicity and scalability, has been repeatedly proven in several large scale industrial case studies. CodeCrawler is implemented on top of Moose. Moose is a language independent reengineering environment written in Smalltalk. It is based on ....

LANZA M., "Combining Metrics and Graphs for Object Oriented Reverse Engineering ", Diploma thesis, University of Bern, Oct. 1999.

....classes using metrics As we have previously seen we use two dimensional boxes to represent classes. We can enrich this representation using metrics: the width and height of the boxes reflect metric measurements of the classes, as we see in Figure 3. This approach has been presented in [LAN 99] and [DEM 99a] In the visualization presented in this paper we visualize classes and use the metrics number of methods (NOM) for the width and number of attributes (NOA) for the height, although in our tool we can choose other metrics. In the case of the differential metrics we just subtract ....

....CodeCrawler and Moose CodeCrawler is the tool used to generate the views presented in this paper. CodeCrawler supports reverse engineering through the combination of metrics and soft The Evolution Matrix 145 Figure 10. The Difference Evolution Matrix of Supremo ware visualization [LAN 99, DEM 99a, DUC 01] Its power and flexibility, based on simplicity and scalability, has been repeatedly proven in several large scale industrial case studies. CodeCrawler is implemented on top of Moose. Moose is a language independent reengineering environment written in Smalltalk. It is based on ....

LANZA M., "Combining Metrics and Graphs for Object Oriented Reverse Engineering ", Diploma thesis, University of Bern, Oct. 1999.

....reengineering tools built on top of Moose. To enable tools to interact, Moose provides a simple tool registration and lookup mechanism. At this point in time the following tools have been developed: CodeCrawler supports reverse engineering through the combination of metrics and visualization [Lan99] [DDL99] see figure 5) Through simple visualisations which make extensive use of metrics, it enables the user to gain insights in large systems in a short time. CodeCrawler is a tool which works best when we approach a new system and need quick insights to get information on how to proceed. ....

Michele Lanza. Combining metrics and graphs for object oriented reverse engineering. Diploma thesis, University of Bern, October 1999.

....graphical representation of methods and attributes using metrics. 2.2.1 Size and Shape of Methods and Attributes We use the width and height of the boxes to reflect metric measurements of the entities which are represented by the boxes, as we see in Figure 2. This approach has been presented in [14] and [3] For the method boxes we use the metric lines of code (LOC = number of non blank lines in a method body) for the height and the number of invocations (NI = number of static call sites) for the width [16, 9] For the attribute boxes we use the metrics number of direct accesses from within ....

.... body) for the height and the number of invocations (NI = number of static call sites) for the width [16, 9] For the attribute boxes we use the metrics number of direct accesses from within the class (NLA) for the width and number of direct accesses from outside of the class (NGA) for the height [14]. Note that the total number of accesses on an attribute is the sum of NGA and NLA. For further explanations on the metrics please refer to [14] 2.2.2 The Use of Colors in a Class Blueprint We make use of colors to display supplementary information in a class blueprint. In Table 1 we present a ....

[Article contains additional citation context not shown here]

M. Lanza. Combining metrics and graphs for object oriented reverse engineering. Diploma thesis, University of Bern, Oct. 1999.

....on their visualization within the evolution matrix. 2.1 Visualizing Classes using Metrics We use two dimensional boxes to represent classes and use the width and height of the boxes to reflect the metric measurements of the classes, as we see in Figure 1. This approach has been presented in [13] and [2] In the evolution matrix discussed in this paper we visualize classes and therefore use the metrics number of methods (NOM) for the width and number of instance variables (NIV) for the height, although in our tool we can choose other metrics. Width Metric Height Metric CLASS Figure ....

....classes only two are persistent, i.e. the whole system renewed itself nearly completely. 4. CODECRAWLER AND MOOSE CodeCrawler is the tool used to generate the views presented in this paper. CodeCrawler supports reverse engineering through the combination of metrics and software visualization [13, 2, 4]. Its power and flexibility, based on simplicity and scalability, has been repeatedly proven in several large scale industrial case studies. CodeCrawler is implemented on top of Moose. Moose is a language independent reengineering environment written in Smalltalk. It is based on the FAMIX ....

M. Lanza. Combining metrics and graphs for object oriented reverse engineering. Diploma thesis, University of Bern, Oct. 1999.

....graphical representation of methods and attributes using metrics. 2.2.1 Size and Shape of Methods and Attributes We use the width and height of the boxes to reflect metric measurements of the entities which are represented by the boxes, as we see in Figure 2. This approach has been presented in [18] and [4] For the method boxes we use the metric lines of code (LOC = number of non blank lines in a method body) for the height and the number of invocations (NI = number of static call sites) for the width [19, 13] For the attribute boxes we use the metrics number of direct accesses from within ....

.... body) for the height and the number of invocations (NI = number of static call sites) for the width [19, 13] For the attribute boxes we use the metrics number of direct accesses from within the class (NLA) for the width and number of direct accesses from outside of the class (NGA) for the height [18]. Note that the total number of accesses on an attribute is the sum of NGA and NLA. For further explanations on the metrics please refer to [18] 2.2.2 The Use of Colors in a Class Blueprint We make use of colors to display supplementary information in a class blueprint. In Table 1 we present a ....

[Article contains additional citation context not shown here]

M. Lanza. Combining metrics and graphs for object oriented reverse engineering. Diploma thesis, University of Bern, Oct. 1999.

....Graph types. We selected a small set of graph types for their simplicity and easy layout algorithm. Table 2 lists their properties and below is a brief description of each of them. For a more detailed description of the graphs and a discussion of their advantages and limitations, we refer to [21]. Tree. Positions all entities according to some hierarchical relationship. Most often used to display class hierarchies. Layout variations are possible by aligning nodes relative to the position of their parent (left, center or right) example Figure 3] Correlation. Positions entities in an ....

....as the height of a node and then positions nodes one besides the other in a long row. Is used to detect exceptional cases for metrics that usually correlate, because the stapling effect will normally result in a steady inclining staircase, yet exceptions will break the steady inclination. see [21]] Confrontation. Visualises two different kinds of entities and the relationships between them. Used mainly for analysing access patterns between attributes and methods. The two kinds of entities are positioned in two separate rows and then edges are drawn to represent the relationships. Layout ....

[Article contains additional citation context not shown here]

M. Lanza. Combining metrics and graphs for object oriented reverse engineering. Master's thesis, University of Bern, 1999.

....displayed by a tree where the size of its nodes reflects metric values of the classes they represent. This way, the graphs can be enriched by more than 30 metrics producing a huge number of possibilities. As only a subset of the enriched graphs is meaningful, we identified so called useful graphs [LAN 99] The Need for a Methodology. However, despite the finding of those useful graphs, there is a definite need for a methodology that guides the reverse engineer depending on the state of his reverse engineering. A methodology should state in a goal oriented way which graph to apply depending on ....

....Figure 1 are incorporated into the graph. 2.5. Graph Types We selected a small set of graph types for their simplicity and simple layout algorithms. We list below the essential ones. For a more detailed description of the graphs and a discussion of their advantages and limitations, we refer to [LAN 99] 6 Techniques et sciences informatiques. Volume X n Name Layout Metrics Sort Scope System Complexity Inheritance Tree NIV, NOM, LOC, NO Full System Gives an overview based on the inheritance hierarchies of a whole system. It gives clues on the complexity and structure of the ....

[Article contains additional citation context not shown here]

LANZA M., Combining Metrics And Graphs for Object Oriented Reverse Engineering . Master's thesis, University of Bern, 1999.

....central repository and provides services such as metric computation and refactorings to the reengineering tools built on top of MOOSE. At this point in time the following tools have been developed: # CODECRAWLER supports reverse engineering through the combination of metrics and visualization [11, 3] (see Figure 4) Through simple visualizations which make extensive use of metrics, it enables the user to gain insights in large systems in a short time. CODE CRAWLER is a tool which works best when we approach a new system and need quick insights to get information on how to proceed. ....

M. Lanza. Combining Metrics and Graphs for Object Oriented Reverse Engineering. Diploma thesis, University of Bern, Oct. 1999.

....the Refactoring Browser provides us with a quick understanding of the application without having to dive into the details of the system. Due to the limited space of this paper we could not show applications of all graph layouts and metrics, nor could we include more useful graphs. Please consult [18] for a more complete list. ffl Fast overview. The tree overview helped us to identify and qualify the main hierarchies: the graphical navigation and tool classes like BrowserNavigator, the refactoring classes, the abstract syntax tree representation, the parser and the scanner. More than just ....

M. Lanza. Combining metrics and graphs for object oriented reverse engineering. Master's thesis, University of Bern, 1999.

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