This paper reports on the design and evaluation of the generic editor GenEd for visual notations. The term ``generic'' refers to the characteristic that users can specify domain-specific syntax and semantics for visual notations. We think of formal notations as a very general notion. Examples for visual notations can be found in areas such as music, geography, mathematics, chemistry, etc. Visual programming languages are viewed as a subset of formal notations.
GenEd is intended for supporting the design and analysis of visual notations or visual programming languages. Usually the design of new visual notations is an exploratory approach. Our experience with the design of formal semantics [1] for a completely visual programming language, Pictorial Janus [2,3], has strongly motivated the approach presented in this paper. We argue that designers of notations should be offered an almost free-form, purely declarative style for specifications with immediate support for analysis and reasoning about specifications and for verification through parsing of example drawings. GenEd's parser can operate in two modes. The incremental mode validates drawings after every relevant modification and reports parsing successes and errors to the user. The second mode performs parsing only when demanded by the user.
Our approach is based on a fully formalized theory about visual notations that can be defined from ``first principles'' by starting with point-sets and topology. GenEd's specification language and its reasoning capabilities depend on description logic theory.
This paper makes several contributions to visual languages. It demonstrates that our theory can be applied to a variety of different visual notations, ranging from simple diagrams, over petri nets, to sophisticated visual programming languages. The latter was extensively described in [1]. It also introduces a new full-featured graphic editor based on our formal theory. This theory was introduced in [1] and is extensively presented and related to visual language theory in [4]. Our approach is in contrast to other approaches that use only syntax specifications or favor a generative solution, i.e. they create specialized editors for particular visual languages.
The next section describes our theoretical foundation in more detail. Then follows the main section explaining GenEd's user interface and presenting three example applications. The last sections discuss related work and future research.