The KIELER Lightweight Diagrams (KLighD) project aims at offering transient lightweight representations of models or parts of them, without incorporating complex editing facilities like graphical editors.
Instead graphical or textual representations are to be synthesized from a chosen fraction of a model base and dismissed if they are not needed anymore. This way the Model-View-Controller paradigm (MVC) shall be established at the users' front end of modeling tools. Although KLighD is intended to address graphical as well as textual transient views, the graphical ones are currently in the focus. The automatic arrangement of those views (macro layout) is contributed by KIML.
Feel free to watch the following videos illustrating the idea and possibilities of transient views of models:
For details on the objectives of our KLighD framework refer to the above mentioned publication Just Model! – Putting Automatic Synthesis of Node-Link-Diagrams into Practice (Sec. I, III, & V) first.
In order to build up your own diagrams by means of KLighD we suggest to install some of our provided features from our update site http://rtsys.informatik.uni-kiel.de/~kieler/updatesite/release_pragmatics_0.10.0:
dotis very popular for simple graphs with edge labels. Note that employing Graphviz requires to also install the native library, see http://www.graphviz.org/
planarizationis the recommended algorithm for arranging UML Class Diagrams. The required OGDF Server Binaries will be automatically installed, too!
In order to let the framework draw your desired diagrams for representing given data we need to tell KLighD how obtain nodes, edges, and hierarchy levels from the provided data. To this end we need to implement a so called diagram synthesis. Getting an initial stub of such an implementation is most easiest done by means of our KLighD project wizard (if the entry is not visible, switch the perspective or close and reopen it, or simply choose Other... → KIELER Light-weight diagrams):
Provide the source model type, e.g. via the 'Browse' button. Note that the project containing that class must exist in the workspace or at least any other workspace project must have this class in its class path, e.g. via a bundle dependency. Otherwise it won't be offered by the list of available classes. Of course you might alter the project's name, the name of the diagram synthesis implementation and its containing package. Finally choose your favorite implementation language - we like Xtend very much for this purpose!
You will get the following implementation template, it is already registered at KLighD's dedicated extension point (see the project's plugin.xml).
Before starting with the implementation let us point you at some hints on using Xtend!
createNode()) for each element comprised by the given instance
model (usually called the business, domain, or semantic model) of your data type, here
EPackage. Add those nodes to the children of
root. Create a
KEdge for each relation or link to be contained in your diagram. Similarly to
createNode() there is
createEdge() for that purpose. Set source and target node of those edges accordingly, this will implicitly add the edges to the diagram (via EMF's
EOpposite mechanism). You can reveal an already created node representing a certain domain element by calling domainElement.
getNode(), or simply domainElement.
node. Have a look at UML2UseCaseDiagramSynthesis, which is part of our examples project, on how to do that.
Finally attach figure specifications to the nodes and edges, most conveniently by means of the extension methods provided by the
...Extension classes - see the use case example on that, too. A documentation of those methods in form of a list of all extension methods sorted by the type they can be used with can be found here. (And yes, there is still some lack of documentation...)