Authors: Düdder, Boris
Title: Automatic synthesis of component & connector software architectures with bounded combinatory logic
Language (ISO): en
Abstract: Combinatory logic synthesis is a new type-based approach towards automatic synthesis of software from components in a repository. In this thesis we show how the type-based approach can naturally be used to exploit taxonomic conceptual structures in software architectures and component repositories to enable automatic composition and configuration of components, and also code generation, by associating taxonomic concepts to architectural building blocks such as, in particular, software connectors. Components of a repository are exposed for synthesis as typed combinators, where intersection types are used to represent concepts that specify intended usage and functionality of a component. An algorithm for solving the type inhabitation problem in combinatory logic - does there exist a composition of combinators with a given type? - is then used to automate the retrieval, composition, and configuration of suitable building blocks with respect to a goal specification. Since type inhabitation has high computational complexity, heuristic optimizations for the inhabitation algorithm are essential for making the approach practical. We discuss particularly important (theoretical and pragmatic) optimization strategies and evaluate them by experiments. Furthermore, we apply this synthesis approach to define a method for software connector synthesis for realistic software architectures based on a type theoretic model. We conduct experiments with a rapid prototyping tool that employs this method on complex concrete ERP- and e-Commerce-systems and discuss the results.
Subject Headings: Logic
Software
Engineering
Architecture
Concurrency
Theorem prover
Synthesis
URI: http://hdl.handle.net/2003/33659
http://dx.doi.org/10.17877/DE290R-6528
Issue Date: 2014
Appears in Collections:LS 14 Software Engineering mit dem Schwerpunkt Entwicklung evolutionsfähiger Software und Systeme

Files in This Item:
File Description SizeFormat 
Dissertation.pdfDNB7.53 MBAdobe PDFView/Open


This item is protected by original copyright



All resources in the repository are protected by copyright.