Transactional Intent

Data state in a data management system such as a database is the result of the transactions performed on that data management system. Approaches such as single-message transactions and field calls [Gray1993] come closer than before/after values to expressing the intent of a transaction, the semantic transformation that should be performed on the data state even if that state is different than what was previously read. But intent is an even higher-level semantic description. This paper illustrates the use of intent-based transactions and processes in several applications, and describes the benefits from exploiting transactional intent. We provide an application framework for intent, and discuss some advanced aspects of intent, including its relationship to apologyoriented computing [Helland2007].