David Cok raised an issue (on the jmlspecs-interest mailing list, May 18, 2013), which can be paraphrased as follows:
It seems that the syntax permits more than one represents clause for a given model field. This raises several questions:
Some thoughts by Gary Leavens:
1.In the interest of simplicity, there doesn't seem to be a good reason to allow more than one non-redundant represents clause for a given model field in a type. That decision would avoid the need to answer question 2, and also allow the most common usage of represents clauses.
3. It is convenient sometimes to allow overriding of represents clauses in subtypes, particularly to give a more efficient or more clear implementation of a model field.