Successor ML is a collection of proposed language extensions to Standard ML. A number of these extensions address pattern matching; including adding richer record patterns, or-patterns, and pattern guards. Pattern guards in Successor ML are more general than those found in other languages, which raises some interesting implementation issues.

This paper describes the approach to pattern guards that we are developing as part of an effort to add Successor ML features to the Standard ML of New Jersey system. We describe how our approach can be used in either back-tracking or decision-tree implementations of pattern matching.