The b ow* Hive is a collection of Eclipse plug-ins that can be integrated into various Eclipse-based modeling tools. It adds several features that are useful in particular for model analysis and model exchange. Most of these features put the principle of end user-programming into practice: Modeling tool users can add their own add-ons at runtime without having to re-compile the original modeling tool. We hope that this will be useful in particular for researchers. They can extend a modeling tool with own model analysis features without having to touch the sources of the modeling tool.
The b ow* Toolbox [
While in the rst years, our work concentrated on our own tool, we soon realized that this idea could be useful for other graphical modeling tools as well. Therefore we refactored our code which resulted in Eclipse plug-ins that can Copyright c 2015 for this paper by its authors. Copying permitted for private and academic purposes be used with other modelling tools based on Eclipse using EMF and GMF or Graphiti. As a result, modellers and researchers can add their own extensions to those tools. 2 2.1
We provide two ways for adding a transformation into or from an external le format: either by using XSLT or by using a transformation language which is based on the template language Apache Velocity. A new export/import can be added by creating a transformation script le, another le de ning the transformation metadata (for instance, name or description) and placing both into a subdirectory of the modelling tool's workspace. 2.2
For model analysis purposes, it is often desirable to add custom-de ned attributes to model elements (for example information about cost and duration to an activity or a probability to a control- ow edge). Unfortunately, this is not always supported by the original modelling tool. We created a new Eclipse view called \Attribute View" which allows to add arbitrary attributes to model elements. An attribute can be a numerical or textual value or an URL. In the latter case, the resource located by the URL can be opened from the Attribute View. This makes it possible to attach les or web sites with additional information to any node or vertex in the model. Additionally, a \Filter View" allows to show only those model elements whose attributes have a certain value (for example all activities where the attribute \cost" is larger than a given value).
It is possible to de ne that the shapes of the diagram should be decorated
with icons when the value of a user-de ned attribute is within a given range.
For this purpose, the icon graphics and a map from attribute values to icons
to be shown has to be added to a subfolder of the modeling tool. Using this
mechanism, we were able to include the EPC model variant for Risk-aware
Event-Driven Process Chains [
For improving the model quality, it can be desirable to give the modeller feedback
about possible modelling problems or detected anti-patterns at modelling time.1
For this purpose, we provide a mechanism to de ne checking rules in three
1 The reader who is interested in this, might alternatively consider to use the EMF
Refactor project, see www.eclipse.org/emf-refactor [
Icon languages: the Epsilon Validation language (EVL), openArchitectureWare Check and Prolog. For rules expressed in EVL, it is also possible to de ne so-called \Quick-Fixes", i.e. model transformations that correct a problem.
The b ow* Hive provides the functionality for adding user-de ned validation rules, for selecting the rules that should be applied, for the internationalisation of problem messages and for changing the problem messages according to the users' preferences. Fig. 2 shows how validation rules have been integrated into openOME, a tool for goal-oriented requirements modelling. 2.4
Add-ons The framework allows to de ne \add-ons". An add-on can be described as a sequence of steps which de nes how to... 1. Extract the model data and translate it into the input format that is required by an external program. 2. Call the external program or a chain of such programs with speci ed parameters. 3. Read the result data of the program(s). 4. Show the result by changing one or more views in the modeling tool.
For each of those steps, our framework already provides out-of-the-box components which just need to be invoked in the right order and with the right parameters.
Fig. 3 shows possible executions of an add-on. As illustrated, the components are combined using a pipes-and- lters architecture.
The called program(s) deliver the result by writing either into a le or to the standard output stream of the process. To update a view of the model within the modeling tool, the rightmost component in Fig. 3 a) requires an input in form of prede ned commands. If the external program does not provide the output in the required format, we have to call another program as an adapter. For showing the results of the external program(s), it is possible to use components that can do the following actions: { Printing information into the Eclipse Console View { Adding information about an error, warning or information Eclipse's Problems
View { Changing the label or the graphical attributes (e.g. size, color, font-style) of model elements { Changing the user-de ned attributes of model elements (in addition, this can result in decorating a model element with icons) (as shown in Fig. 1(b)) { Showing a generated HTML page
With the help of the add-on mechanism, extensions in great variety have been
developed. There are add-ons for model-checking, quality-of-service calculation for
business processes, business process simulation [
The b ow* Hive is already integrated into our own EPC modelling tool b ow*
Toolbox (http://b ow.org), into UPROM [
The source code can be obtained from https://github.com/b owtoolbox/app where all plug-ins named org.b ow.toolbox.hive.* are part of the tool-independent b ow* Hive. With the help of these plug-ins, modeling tool developers can include the functionality described in this article into their own Eclipse-based modeling tool. The easiest way for a user to pro t from the features provided by the b ow* Hive in the modelling language of his or her choice is to download the most recent version of the b ow* Toolbox from the web site mentioned above and to use Eclipse's update mechanism for adding support for other modelling languages.
We are looking forward to reports from people who made use of the b ow* Hive into their tools, developed le exports, imports or add-ons using our mechanism or improved the source-code of the b ow* Hive.
Any questions related to our plug-ins are welcome to bflow@bflow.org.