Catacomb Design and Implementation

Catacomb is intended to be easily separable into a variety of different components covering areas including model description, on-the-fly graphical interface construction, model construction, model publishing, computation of the behavior of models, and data display.

Contents

• A little background showing what the perceived problems are that Catacomb is trying to address, and what a solution might look like.

• Declarative model description is at the heart of Catacomb. Superficially, it just makes it easy to write generic user interfaces, but it also turns out to be a very efficient way to organize models and indeed many other sorts of data.

• The Design Summary shows how Catacomb works, how models are described and constructed, and what goes on behind the scenes and how calculations are structured.

• On-the-fly user interface generation relies on the declarative model approach. Because only a small set of data structures is used, Catacomb is able to provide user interface tools giving complete coverage of all models consistent with the allowed structures.

• Event propagation is provided within the model description tree, independent of the events in the windowing toolkit. Under normal operation, any parameter change in the model percolates up through the hierarchy and along references. This can be used to ensure consistency of the model and any derived properties.

• Running Models is the main goal of Catacomb. Each of its model description frameworks has a corresponding family of model execution classes. These can be instantiated from the model description ore by other means.Thereafter, they run independently of the description machinery.

• Publishing models and data. Publishing models is straightforward: the standard file format starts with a small section of HTML header which includes javascript links to start up the Catacomb user interface and load the model. There is an applet builder (which writes a declarative description of what goes in the applet) to define compound applets for viewing in a single window. For publishing data, Catacomb provides a way to describe how a particular class of data could be documented, and then tools to build documents consistent with this format. The first can be used by domain experts, or anyone who finds the available formats inadequate. The second is for documenting particular chunks of data before publishing them. Catacomb can also make simple websites, indexes, and alert interested servers to the presence of the data so the site can be spidered.

• Interoperability. Catacomb is intended to be interoperable with other systems, including Neuron and Neosim. Here is how it is done.