WG211/M11Sloane

Profile-based Analysis and Abstraction of Attribute Evaluation by Tony Sloane

Attribute grammars enable complex algorithms to be defined on tree and graph structures using simple local equations. A full understanding of the algorithms can be hard to obtain from the individual equations because it requires an understanding of how information propagates through the whole structure.

We use profiling in our Kiama language processing library to collect information about attribute evaluation and summarise it along dimensions including attribute dependencies, values, and locations or types of the nodes at which attributes are evaluated. Attribute grammar developers can use built-in dimensions to get a general view of the operation of their equations or add their own dimensions to summarise the information in a problem-specific way.