Daniel Fernandez-Lanvin

The Runtime Performance of invokedynamic: An Evaluation with a Java Library

The Java 7 platform includes the invokedynamic opcode in its virtual machine, a feature that lets programmers define-and dynamically change-the linkage of method call sites, thereby maintaining platform optimizations. A comprehensive evaluation of a new librarys performance includes a description of how to optimize real Java applications.

Measuring the role of age in user performance during interaction with computers

Abstract The influence of aging on computer interaction has been widely analyzed in human–computer interaction research literature. Despite this, there are no age-based user maps that could support the user-interface customization. Studying the specific needs and constraints of these groups is crucial in order to adapt a user interface to the user’s interaction requirements. This work studies the performance of a sample of participants on three different basic tasks (pointing, dragging and dropping, and text selection) and the influence of age for each of them. It is concluded that this influence differs between specific activities. A group profile map that can support automatic classification in the future has been obtained.

Implementing Automatic Error Recovery Support for Rich Web Clients

The way developers usually implement recoverability in object oriented applications is by delegating the backward error recovery logic to the ever-present database transactions, discarding the in-memory object graphwhen something goes wrong and reconstructing its previous version from the repository. This is not elegant from the point of view of design, but a cheap and efficient way to recover the system from an error. In some architectures like RIA, the domain logic is managed in the client without that resource, and the error prone and complex recoverability logic must be implemented manually, leading to a tangled and obfuscated code. An automatic recovery mechanism is adapted to that architecture by means of a JavaScript implementation. We developed several benchmarks representing common scenarios to measure the benefits and costs of this approach, evidencing the feasibility of the automatic recovery logic but an unexpected overhead of the chosen implementation of AOP for JavaScript.

Web Design Attributes in Building User Trust, Satisfaction, and Loyalty for a High Uncertainty Avoidance Culture

In this study, we attempt to evaluate the user preferences for web design attributes (i.e., typography, color, content quality, interactivity, and navigation) to determine the trust, satisfaction, and loyalty for uncertainty avoidance cultures. Content quality and navigation have been observed as strong factors in building user trust with e-commerce websites. In contrast, interactivity, color, and typography have been observed as strong determinants of user satisfaction. The most relevant and interesting finding is related to typography, which has been rarely discussed in e-commerce literature. A questionnaire was designed to collect data to corroborate the proposed model and hypotheses. Furthermore, the partial least-squares method was adopted to analyze the collected data from the students who participated in the test (

The dimension of age and gender as user model demographic factors for automatic personalization in e-commerce sites

n

Domain model slicing and constraint classification for local validation on rich clients

= 558). Finally, the results of this study provide strong support to the proposed model and hypotheses. Therefore, all the web design attributes were observed as important design features to develop user trust and satisfaction for uncertainty avoidance cultures. Although both factors seem to be relevant, the relationship between trust and loyalty was observed to be stronger than between satisfaction and loyalty; thus, trust seems to be a stronger determinant of loyalty for risk/high uncertainty avoidance cultures.

Integrating automatic backward error recovery in asynchronous rich clients

Abstract Personalization in e-commerce increases sales by improving customer perception of site quality. However, some demographic data about customers (crucial for the success of the personalization process) not always can be obtained explicitly, as is the case of anonymous web site visitors. The paper describes a user study focused on determining whether it would be possible to categorize the age and gender of individual visitors of a web site through the automatic analysis of their behavior. Three tasks commonly found in e-commerce sites (Point & Click, Drag & Drop and Item Selection) were tested by 592 volunteers and their performance was analyzed using several different statistical methods. The study found consistencies in the execution times of individuals across the different tasks and revealed that age and gender are sufficiently determining factors to support an automatic profiling. Results also showed that relevant information about gender and age can be extracted separately through the individual analysis of each one of the mentioned interaction tasks.

OCL for rich domain models implementation: An incremental aspect based solution

Abstract Web-based rich client applications have emerged as a solid and popular approach in both web and native applications. Their capability to manage their own domain model and locally verify constraints provides a more responsive and robust user experience. This local model is often a subset of the applications global domain model (GDM) that is managed on the server. Both ends should always manage their entities, relationships and constraints consistently between them. Designing such client model manually implies identifying the GDM domain elements and constraints that should also be present on the client and adapting each one of them if needed. This is a complex and error-prone task, and any additional modification to the server model requires reviewing the client side. In our opinion, all the information needed for automating the client model generation can be derived from the GDM and the set of entities involved in the client functionality. This work includes a formal description of a method that, from that initial information, combines model slicing and constraint analysis techniques to create the client domain model, and classifies the constraints according to their server independency.

Multi sensor system for pedestrian tracking and activity recognition in indoor environments

Rich Web Clients allow developers to manage data locally and update it from the server by means of asynchronous requests, thus providing more interactive interfaces and an improved user experience. On the other hand, they face concerning challenges regarding error management. When there is a need to update the local data through multiple asynchronous requests and it is required that all them succeed, an error on a single call can lead to having incorrect information shown to the user. Consequently, developers need to explicitly implement proper recovery mechanisms, a task that most of times is complex and highly error prone, leading to tangled code and harder maintenance, especially in an asynchronous environment. These problems could be lessened through automatic error recovery techniques, but the existing state of the art for Rich Web Client development does not support recovery from asynchronous scenarios. To cope with this problem we extended the existing error recovery technique of Reconstructors, adding to it the capability of recovering the state in the presence of several asynchronous requests. We applied this technology in a widely used open source project for rendering interactive charts, ChartJs, thus allowing the developer to effortlessly guarantee that the data shown to the user, even when it results from multiple asynchronous requests, is never inconsistent. We compare our proposal to other solutions using state of the art approaches and verify that by using Reconstructors the overall implementation requires 39.16% less lines of code, and 5.66 times less lines of code are dedicated specifically to error management, while avoiding code tangling completely. Execution time showed by reconstructors is between 5.2% and 9.3% slower than other solutions, a cost we believe is worth its benefits, and feasible for using these techniques in real world client applications.

Cost estimation in software engineering projects with web components development

Object Constraint Language (OCL) can be used to express domain model constraints. Those related to a single object are easy to implement. However, when a constraint depends on the state of more than one object (domain and class constraints) the problem turns much more complicated. Developers must deal with several difficulties: how to write the invariant check, when to execute the constraint verification, over what objects and what to do in case of a constraint violation. Things are harder if we add feasible performance as requirement. We propose a tool that combines incremental OCL processing, with translation into aspect code and execution inside an atomicity execution context. The output is aspect code, ready to be integrated with business code that checks all the invariants efficiently at the end of the atomic operation.