|
|
C-language code generator for SOFA 2
Ježek, Lukáš ; Bureš, Tomáš (advisor) ; Bulej, Lubomír (referee)
SOFA 2 is a component system employing hierarchically composed components. It provides ADL-based design, behavior specification using behavior protocols, dynamic reconfiguration of the components, and modeling of the component communication by software connectors. This allows seamless and transparent distribution of component applications. The connectors can be automatically generated, SOFA 2 contains Java connector generator allowing to connect components with Java interfaces. The aim of this thesis is to implement C code generator and integrate it into the current SOFA 2 connector generator framework, so that C connectors can be automatically generated and thus components written in C language can be transparently connected in distributed environment. The proposed C code generator is based on the concept of template transformation, where templates containing mixture of C code and a scripting Domain Specific Language are transformed to a pure C code. Strategic term rewriting method provided by Stratego/XT framework is used for evaluation of the scripts within the templates.
|
|
|
Infrastructure for Deployment of Heterogeneous Component-based Applications
Šafrata, Pavel ; Bulej, Lubomír (advisor) ; Hnětynka, Petr (referee)
Deployment is a process which involves all actions performed with an application after it is released. Traditionally, deployment has been addressed for each component model separately (if at all), even though most of the concepts are the same. The Deployment and Configuration of Component-based Applications Specification released by OMG proposes a unified approach that can be tailored to different component models. This thesis focuses on the execution phases of the deployment process. It presents a generic deployment runtime based on the OMG specification. The main objective is to elaborate support for multiple component models and subsequently support for heterogeneous applications consisting of components implemented in different component models. This has been achieved through a system of extensions which allows isolating component model specifics from the runtime. Even though the OMG specification was not originally intended to support heterogeneous applications, the implementation deviates from it only in a few points. In all such cases, the thesis presents an analysis of the situation and rationale for the deviation.
|
|
|
Mass transport routing
Bečka, Michal ; Bulej, Lubomír (advisor) ; Yaghob, Jakub (referee)
A journey planning tool accessible from a mobile device provides travel information on the move. This work studies various aspects of journey planners in order to develop a solution for this task. We analyze the connection with phone networks and the Internet for possible choices to place the tool. Then we study the development environment consisting of operating systems programming languages. Another part covers possible parameters for the journey search. Then we investigate the source and form of input traffic data along with possible algorithms for this problem. Based on this analysis we design an application, making decisions from the analyzed areas. The design is applied for the following implementation and the resulting program we describe and measure. The purpose of this work is to develop a working application for mobile devices capable of public transit journey search, and to show what development of such application entails.
|
|
|
Simulink Block Library for LEGO NXT
Škoda, Dominik ; Bureš, Tomáš (advisor) ; Bulej, Lubomír (referee)
Simulink Block Library for LEGO NXT Abstract Dominik Škoda July 30, 2014 The goal of this work is to create a support for the LEGO NXT platform in Simulink development environment. Such support of the target platform already exists, but it suffers from several disadvantages. At first it is provided exclusively for Windows operating systems, and the implementation is closed, therefor neither extensible nor customizable. The main premise of this project is the support of Linux operating systems. The project is also opened to ensure the extensibility and customizability. The model-driven development of systems for the LEGO NXT platform using this project comprises the model testing in a simulation and code generation in Simulink environment by using its standard tools, and deployment of completed programs to target devices. The systems generated with the help of this project are categorized as real-time systems.
|
|
|
Deployment Planner for Heterogeneous Component-based Applications
Urban, Jaroslav ; Bulej, Lubomír (advisor) ; Hnětynka, Petr (referee)
Deployment of component applications is the process of making them available for further use by clients. The OMG Deployment & Conguration Speci cation aims at creating a uni ed deployment process independent of component models. We use the OMG speci cation to create a deployment framework able to deploy heterogeneous component applications which are implemented using multiple component models. This work focuses on the planning phase of the speci cation, which selects computer resources for components with respect to their requirements. We have implemented the planning phase via a graphical user interface utilizing an automated planning algorithm which assists the user in creating a valid planning. The planning tools are extensible to support additional technologies, component models and more advanced planning heuristics.
|
|
|
Dynamic linking, debugging and tracing in HelenOS
Svoboda, Jiří ; Děcký, Martin (advisor) ; Bulej, Lubomír (referee)
HelenOS is an operating system that originated as a software project at the Faculty of Mathematics and Physics. So far it lacks support for dynamically linked libraries as well as support for process tracing and debugging. Dynamically linked libraries enable developing individual parts of large software systems independently and linking them later together without recompilation. The linking is carried out at load-time or run-time by the dynamic linker. The linker must nd all libraries used by the program, map them into memory and relocate them. Then it must resolve external (symbolic) references between the program and libraries. A debugger and a system-call tracer are essential development tools. They use a special system interface for their operation enabling them to suspend an application when certain events occur (such as a breakpoint or a trap). Then they may examine or change the application's memory contents and resume its execution. The main goal of this thesis is to implement support for dynamically linked libraries in HelenOS, namely the dynamic linker, and also a system API for debugging and tracing processes, including a demo application.
|
|
|
Networking and TCP/IP stack for HelenOS system
Mejdrech, Lukáš ; Děcký, Martin (advisor) ; Bulej, Lubomír (referee)
Within this work we study networking and TCP/IP stack implementation. The main interest is directed to the TCP/IP stack design and implementation respecting the microkernel operating system concept. The practical part was a development of a TCP/IP stack for the microkernel operating system HelenOS. First, we describe the networking and the TCP/IP stack concept in general. The special aspects of the microkernel concept follow. For the practical part, the architecture design and implementation decisions are included.
|
|
|
Connector-based Performance Data Collection for Component Applications
Bulej, Lubomír ; Tůma, Petr (advisor) ; Janeček, Jan (referee) ; Murphy, John (referee)
In this work, we propose a generic approach to collection of performance data for hetero-geneous component-based applications with the aim to provide easier and less costly access to performance data needed for measurement and model-based performance analysis of component applications. The technical foundation for the approach is built on generic solutions to various aspects of performance data collection and is made of three parts. The first part provides a design of a generic measurement infrastructure which handles common performance measurement tasks and allows collecting arbitrary performance data in response to performance events. The second part proposes using architecture-based connectors for instrumentation of component applications and provides a design of a performance instrumentation connector element for use with the measurement infrastructure. The third part proposes integration of connectors into deployment process of component applications which enables deployment and transparent instrumentation of heterogeneous component-based applications.
|
|
| |
|
|
Data Processing in a Generic Benchmarking Environment
Mácha, Radek ; Podzimek, Andrej (advisor) ; Bulej, Lubomír (referee)
Title: Data Processing in a Generic Benchmarking Environment Author: Radek Mácha Department: Department of Distributed and Dependable Systems Supervisor: RNDr. Andrej Podzimek, Department of Distributed and Dependable Systems Abstract: In September 2013, at the Charles University in Prague, Faculty of Mathematics and Physics, a software project implementing the third incarnation of a generic bench- marking environment aimed at performance evaluation of networked applications was presented: EverBEEN. Despite significant advancements achieved by this incarnation in both reliability and ease of use, EverBEEN still came out somewhat wanting in terms of commercial usability. One of its major shortcomings was the absence of a standardized way of data extraction and processing. The want of such means in EverBEEN laid foundation to the central question of this thesis: How to extract and process data from a framework like EverBEEN, with no prior knowledge of the structure of said data? Albeit centered on the creation of a common, reusable data extraction and aggregation codebase for said framework, this thesis also strives to analyze means of automating EverBEEN control-flow and incorporating the framework, and its data processing, into continuous integration. Keywords: performance evaluation, data processing,...
|
guest ::
