Existing CASE tools are often rigid and do not support the users' native methodologies. To alleviate this, more flexible and customisable tools called CASE shells are emerging. However, the customisation of those tools is still cumbersome and error-prone, and demands several configuration files that follow a rigid syntax of some metamodelling language(s). In order to make the eustomisation easier, we propose a graphical metamodelling editor, MetaEdit, with which the conceptual structures of the user methodology can be modelled easily using an easy-to-grasp graphical notation. With MetaEdit, methodology models can be ¢onsmacted with less effort and the configuration files for the CASE sheU can be created (semi)automatically. The tool is flexible i.e. its symbols and metamodel are user-definable. In consequence it can be used as a simple CASE shell. MetaEdit is based on the Object-Property-Role-Relationship (OPRR) data model. The paper presents the principles on which the editor is built, describes its operation, and discusses its relations to other research on metamodelling.

No abstract

In spring 1989 we surveyed the experiences of some Finnish companies in methodology modelling (metamodelling) and adaptation of tools and methodologies to each other (methodology adaptation). The companies represented software production, banking, wood and metal industry, and wholesale trade. The study was carried out as a feld study where we interviewed method developers, systems analysts and their supervisors. The goal of the survey was to find out whether there was need for metamodelling or methodology adaptation in general and how this need had been satisfied. The study shows that a little experience had been gained in adapting data dictionaries to methodologies but no such attempts had been made with CASE tools. One reason for this was that few methodological guidelines were extensively employed and supported in organisations. In general no systematic approach had been followed in the adaptation. We also explored possible causes for encountered difficulties and conditions for successful metamodelling and methodology adaptation. A cmcial success factor was the adaptability and ease of use of the tool. In general the causes for the success are similar to those of information systems development in general. These include sufficient resources and management support, concrete benefits for those who do the actual work and friendly user interfaces.

In this paper we discuss some requirements for future CASE (Computer Aided Software/Systems Engineering) environments. These requirements include increased modifiability and flexibility as well as support for process and agent models. We claim that they can only be addressed by developing more powerful representation and modeling techniques. As a possible basis for modeling various techniques, we outline a general information architecture for a future CASE environment. In addition, we propose primitive types for specifying techniques, the development process, and agent models, and use these types for modeling an example methodology and examine how the requirements are or can be supported in our example.

No abstract