Point-based temporal extensions of SQL and their efficient implementation [chapter]

David Toman
1998 Lecture Notes in Computer Science  
This chapter introduces a new approach to temporal extensions of SQL. In contrast with most of the current proposals we use single time points as references to time, while still achieving efficient query evaluation. The proposed language, SQL/TP, naturally extends the syntax of SQL/92 : it adds a single new data type that represents a linearly ordered universe of time instants. The semantics of the new language extends the standard SQL in the expected way: the new data type behaves identically
more » ... o the existing types. We also eliminate or fix many of the problems connected with defining a precise semantics in interval-based languages. In addition we provide an efficient query evaluation procedure based on a compilation technique that translates SQL/TP queries to SQL/92. Therefore existing off-shelf database systems can be used as back-ends for managing temporal data. 1 Why yet another temporal extension of SQL? After more than a decade of research in the area temporal databases there is still no universal consensus on how temporal features should added to the standard relational model. Instead, there are dozens of mutually incompatible models and proposals. The more practical of these are often based on (often ad-hoc) extensions of existing relational languages, e.g., TQUEL [17] and various temporal extensions of SQL: TSQL2 [18], ATSQL2 [5], and SQL/Wemporal [19], the current proposal of temporal extension of SQL3 to the ISO and ANSI standardization committees. The goal of this chapter is twofold: First, we point out severe problems common to the majority of current proposals, namely representation dependent handling of temporal values and limited temporal dimensionality of the underlying temporal models. We argue that these problems are inherently tied to the use of interval-valued temporal attributes and cause major problems when a precise semantics of the temporal query languages is to be defined. Second, we propose
doi:10.1007/bfb0053704 fatcat:ycv4x5unwza3fpxygyrlnkadzu