The paper describes MPVisualizer (Message Passing Visualizer) a general purpose tool for the debugging of message passing parallel applications; its three components are the trace/replay mechanism, the graphical user interface and the central component, called visualization engine. The engine, which plays the main role during the replay phase, can be used with different message passing environments and different graphical environments. This is a major step to make MPVisualizer a general tool. Additionally, the engine is able to recognize potential race conditions and can be easily re-programmed to detect specific predicates.

A prototype for the visualization of both geometrical and non-geometrical data is described. Special emphasis is given to flat structured or tabular data. Main features of this prototype are: a data independent architecture; filtering of information using degree of interest functions and establishing limits for the variation of one or more data attributes; interactive specification of attribute restrictions combining and and or operators; choice of representations using current scale factor and the result of the degree of interest function. The concept of information classes is introduced to associate a set of attributes with a spatial reference and a graphical representation. Derived information classes are used to combine attribute restrictions with and and or operators.

We describe a platform to build virtual environments inhabited by autonomous virtual humans. We propose an architecture comprising several software components assembled to define a platform. The architecture includes a 3D-modelling software, a rendering engine, a library for simulating rigid body dynamics and an agent development framework.

An object-oriented conceptual model for the debugging of message-passing distributed programs incorporates several debugging facilities: generation of a space-time diagram showing the progression of the execution being studied, detection of race conditions, detection of particular kinds of predicates and representation of causality cones. The focus of this paper is the race detection mechanism. The proposed mechanism comprises two steps: detection of pairs of receive events in the same process potentially involved in race conditions and verification of the legitimacy of the potential race condition. The mechanism relies on the analysis of two arguments, process id and message tag, in receive events and consume events, which are considered as distinct types of occurrences in the conceptual model.

MPVisualizer (Message Passing Visualizer) is a tool for the monitoring and debugging of message passing parallel applications with three components: the trace/replay mechanism, the graphical user interface and a central component, called visualization engine. The engine, which plays the main role during the replay phase, builds an object-oriented model of the application. Taking full advantage of inheritance and polymorphism the tool can be adapted to different message passing environments and different graphical environments, and easily reprogrammed to detect specific predicates. The engine is also prepared to recognize race conditions.

The paper describes a fundamental feature in a prototype for the visualization of both geometrical and non-geometrical data: filtering mechanisms. The information class is a basic concept of this prototype. Each class combines a set of attributes associated to the same spatial reference with a list of graphical representations. Information filtering can be accomplished using the following mechanisms: limiting the variation domain of information class attributes, suppressing the visualization of one or more information classes, using the degree of interest function, selecting specific graphical representations accordingly both to the scale of representation and to the degree of interest function, displaying the data by levels of interest. ° This work was supported by project SARA PRAXIS /2/2.1/TIT/1662/95

Resumo Neste artigo apresenta-se uma plataforma de suporte à geração de cenas animadas com agentes inteligentes. É proposta uma arquitectura que separa o processamento gráfico do processamento inteligente. A plataforma que a concretiza incorpora várias aplicações que são articuladas por dois módulos independentes que comunicam entre si: a camada de processamento gráfico e a camada de processamento inteligente. Palavras Chave: humanos virtuais, ambientes virtuais, animação comportamental.