Jan Scheffczyk

Towards XML version control of office documents

Office applications such as OpenOffice and Microsoft Office are widely used to edit the majority of todays business documents: office documents. Usually, version control systems consider office documents as binary objects, thus severely hindering collaborative work. Since XML has become a de-facto standard for office applications, we focus on versioning office documents by structured XML version control approaches. This enables state-of-the-art version control for office documents.A basic prerequisite to XML version control is a diff algorithm, which detects structural changes between XML documents. In this paper, we evaluate state-of-the-art XML diff algorithms w.r.t. their suitability to OpenOffice XML documents and the future OASIS office document standard. It turns out that, due to the specific XML office format, a careful examination of the diff algorithm characteristics is necessary. Therefore, we identify important features for XML diff approaches to handle office documents. We have implemented a first OpenOffice versioning API that can be used in version control systems as a replacement for line-based or binary diffs, which are currently used.

Formale Konsistenzsicherung in informellen Software-Spezifikationen

Zusammenfassung.Die Spezifikation ist die Grundlage für den Erfolg eines Software-Projekts. Der Praktiker konzentriert sich hier i.d.R. auf die größte Herausforderung: die vollständige und inhaltlich korrekte Erfassung aller Anforderungen an das zu erstellende Software-System. Dass Spezifikationen für große Systeme aus vielen Dokumenten verschiedener Ausprägung in Form und Inhalt bestehen, tritt dabei meist in den Hintergrund. Ihre Konsistenz wird meist mit hohem manuellem Aufwand sichergestellt. Der Erstellung formal korrekter und konsistenter Software-Spezifikationen widmen sich zahlreiche theoretische Arbeiten. In der Praxis sind sie jedoch meist nicht mit der gewohnten Arbeitsweise vereinbar. In diesem Artikel stellen wir einen Mittelweg vor: Wir nutzen die Spezifikationsbausteine von sd&m für die Spezifikation, die als Ergebnisse Dokumente in natürlicher Sprache sowie semi-formale Darstellungen umfasst. Für Spezifikationen, die nach diesen Bausteinen erstellt wurden, definieren wir Konsistenz durch formale zeitbehaftete Konsistenzregeln. Ein von uns entwickeltes Auswertungswerkzeug ermittelt Inkonsistenzen präzise. Unterstützt von einem solchen Werkzeug kann sich der Software-Ingenieur wieder ganz auf das Hauptanliegen der Spezifikation konzentrieren: ihre inhaltliche Korrektheit und Vollständigkeit.Abstract.Specification is the base for the success of a software project. Usually, practitioners concentrate on completely and correctly capturing all requirements for the specified system. In practice, a serious problem often fails to gain attention: Specifications for large systems consist of many documents with heterogeneous structure and content. Their consistency is often achieved by time-consuming manual effort. On the other hand, many theoretical approaches aim at formally correct and consistent software specifications. In practice, however, these approaches fail to be integrated into the every day work of software engineers. This article is located at the boundary between theory and practice: For specifying software we use analysis modules developed at sd&m, the results of which are documents in natural language or semi-formal documents. For a specification using our analysis modules we define consistency via formal temporal consistency rules. Our newly developed tool pinpoints inconsistencies precisely. Supported by our tool, software designers can concentrate on their actual work: the technical correctness and completeness of a specification.

Preservation of digital publications: an OAIS extension and implementation

Over the last decades, the amount of digital documents has increased exponentially. Nevertheless, traditional document engineering methods are applied. Even worse, the long-term preservation issues have been neglected in standard document life cycle implementations.Our digital (cultural) heritage is, therefore, highly endangered by the silent obsolescence of data formats, software and hardware. Severe losses of information already happened. It is high time to implement concrete solutions.Fortunately numerous institutions already target these issues. Moreover, with the OAIS reference model1 a rich standardized conceptual framework is available, which already serves as implementation basis.2This paper discusses an extension to the OAIS reference model and illustrates a prototype implementation of a document life cycle that is enriched by functions for long-term preservation.More precisely, this paper aims to provide first solutions to the following three problem areas:1. Detachment: OAIS defines no functions for the process of detaching digital documents prior to the ingest function. This detachment function is modeled in great detail and implemented for the provision of the so-called OAISs submission information packages (SIP).2. DBMS: OAIS defines a very complex functionality. We show how a standard database management system (DBMS) can support a wide variety of required functionalities in an integrated and homogenous way. Among others OAISs data management, archival storage, and access are supported.3. Metadata: So far, OAIS does not cover any aspects of the metadata generation. Here, we briefly discuss the (semi-)automatic generation of a metadata set.In order to evaluate the feasibility of our approach, we built a first prototype. We carried out our experiments in close cooperation with the Bavarian State Library, Munich, which is engaged in numerous international initiatives dealing with the problem of long-term preservation. Our University Library also supported us by delivering a representative test set of digital publications.3We conclude our paper by presenting some lessons learned from our conceptual work and from our real world experiments.

Pragmatic consistency management in industrial requirements specifications

Industrial requirements specifications suffer from consistency problems, particularly in multi-angular or temporal relationships between different specification results. Current consistency management tools generate too many repairs, lack support for temporal relationships, and are poorly integrated into development processes. In this paper we evaluate our consistency management method on how it improves quality of industrial specifications: we formalize (temporal) consistency rules and generate a few domain-specific repairs for inconsistencies. We demonstrate our method using an example specification. Since the effort for formalization is tunable to specific applications and our prototype shows satisfactory performance, we are confident that our contributions scale to an industrial setting.