NMF Incerator

Incerator is a tool to optimize the creation of dynamic dependency graphs.

When should I use this package?

Incerator supports the following strategies to build dynamic dependency graphs (DDGs):

• Instruction-level incrementalization: Every node in the AST is turned into a DDG node
• Batch: An entire expression is recalculated based on static analysis (e.g., if some property has changed that was used to calculate the result)
• Parameter-promotion: For a given method, the system calculates the minimal amount of model elements from which all changes to a given expression can be obtained. If this is not the case for the parameters of a method (due to cross-references), an optimized method is created that fulfills this condition.
• Tree-extension: Like parameter-promotion, but here the system uses parent model elements to conclude that some root elements are sufficient to capture all changes.

Parameter-promotion and tree extension only work if the models use the NMF modeling foundation.

You can change the strategy to create DDG nodes with a static property, demonstrated by the following line of code:

NotifySystem.DefaultSystem = new PromotionNotifySystem();

Alternatively, Incerator also implements a notify system that records expressions where the decision is necessary and an implementation that decides for each expression separately, based on a configuration. Finally, a tool delivered with this package and accessible using the PowerShell console can optimize a given application for the best configuration using a genetic search algorithm.

Where can I get more information?

There is a documentation website which we try to maintain that contains a few tutorials. There are publications if you want to explore the technical details. Also, please feel free to ask a question or report a bug.