M.H.Bohlen and C. S. Jensen. A Seamless Integration of Time into SQL. TR R-96--2049.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....are valid at the same time. Also, if the user is not interested in the time aspects and does not mention them in a query, the behaviour should be consistent with that of a standard database (these issues are known as upward compatibility, snapshot reducibility, support of legacy applications, etc. [BJ96]) Because a user does not declare and see explicit time attributes, there is no big incentive to offer e.g. time intervals with a number of operations in the style of abstract data types that would then be used like other data types in querying (although some work in this direction exists ....

Bhlen, M.H., and C. Jensen, Seamless Integration of Time into SQL. Dept. of Computer Science, Aalborg University, Technical Report R-96-49, Dec. 1996.

....time instant at which the represented fact holds in the modeled reality. Therefore, tuples are associated with compact encodings of a set of time instants (often called period of validity) The sets of time instants are commonly stored using (finite sets of) intervals [17, 19] bitemporal elements [4, 14], or other fixeddimensional products of intervals (hyper rectangles) The syntax of the chosen encoding then provides a domain of values for temporal attributes in the languages, e.g. pairs of interval endpoints (ATSQL [5] and SQL Temporal [19] use the BEGIN and END keywords to extract the ....

....to compare the timestamps) However, this is only possible if the user knows that the timestamps are encoded by intervals. It also leads to a tension between the syntax of the query languages and their intended semantics: the data model and the semantics of the languages are point based 1 [4, 9], while temporal attributes in the syntax refer to the actual encoding for sets of time instants (e.g. interval endpoints) This conflict leads to several unpleasant surprises when precise semantics needs to be defined. Most importantly, it is easy to show examples of queries whose answers depend ....

Bohlen, M. H., Jensen, C. S. Seamless Integration of Time into SQL. University of Aalborg, http://www.cs.auc.dk/ boehlen/Software/Tiger/atsql.ps.gz, 1996.

....first direction is concerned with temporal extensions to practical query languages such as SQL, e.g. GN93, NA93, Sar93] The issues addressed include space efficient storage, effective implementation techniques, and handling of large amounts of data. This direction of research includes ATSQL [BJ96] an integration of ideas from TSQL2 [Sno95] and ChronoLog [Boh94] The second direction of research is concerned with defining high level query languages with logical semantics, e.g. temporal logic [TC90, GM91, CCT94] The advantages of using logic based query languages come from their ....

.... Gad88, Sno95] The paper is organized as follows: We start with a discussion of the basic framework in Section 2, including the syntax and semantics of temporal logic and ATSQL (in the case of ATSQL we introduce only constructs relevant to the development in this paper; for full description see [BJ96] In Section 3 we give the mapping from temporal logic to ATSQL.We conclude the section with 1 In this paper we use the term period rather than the term interval commonly used in temporal logic because the latter term conflicts with the SQL INTERVALs, which are unanchored durations, such ....

[Article contains additional citation context not shown here]

M. H. Bohlen and C. S. Jensen. A Seamless Integration of Time into SQL. ACM Transactions on Database Systems, 1996.

....are valid at the same time. Also, if the user is not interested in the time aspects and does not mention them in a query, the behavior should be consistent with that of a standard database (these issues are known as upward compatibility, snapshot reducibility, support of legacy applications, etc. [BJ96]) Because a user does not declare and see explicit time attributes, there is no big incentive to offer e.g. time intervals with a number of operations in the style of abstract data types that would then be used like other data types in querying (although some work in this direction exists ....

Böhlen, M.H., and C. Jensen, Seamless Integration of Time into SQL. Dept. of Computer Science, Aalborg University, Technical Report R-96-49, Dec. 1996.

....space requirements: a tuple would have to be repeated for every time instant at which the fact represented by the tuple holds. Instead, compact encodings of sets of time instants (often called periods of validity) associated with a tuple are encoded using intervals [16, 18] bitemporal elements [4, 14], or other fixed dimensional products of intervals (hyperrectangles) The chosen encoding then provides a domain of values for temporal attributes. However, in all the above approaches, the explicit access to the intervalbased encoding of timestamps in temporal relations leads to a tension ....

.... leads to a tension between the syntax and the intended semantics of the languages, specifically: References to time are realized using temporal attributes explicitly referring to intervals (or some other particular encoding of timestamps) The data model and the semantics are point based [4, 9]: the intervals are used merely as compact descriptions of large sets of time instants (even in cases when one of the various duplicate semantics is used) This conflict leads to many unpleasant surprises , especially when multiple temporal dimensions are needed to evaluate a given query (e.g. ....

Bohlen, M. H., Jensen, C. S. Seamless Integration of Time into SQL. University of Aalborg, http://www.cs.auc.dk/ boehlen/Software/Tiger/atsql.ps.gz, 1996.

No context found.

M.H.Bohlen and C. S. Jensen. A Seamless Integration of Time into SQL. TR R-96--2049.

No context found.

Bohlen, M. H. and C. S. Jensen. Seamless Integration of Time into SQL. Technical Report R-962049.

....needed to define modifications of databases with the variable NOW , termed variable databases [10] Later we compare the semantics of modifications on ground and variable databases. Most existing temporal data models support intervals without NOW in both queries and modifications, see, e.g. [6, 26]. We focus on the valid time aspect of the tuples, i.e. when the information recorded by the tuples is true in the miniworld [19] The transaction time aspect, when tuples are current in the database, is a simpler special case because transaction times are maintained by the database management ....

.... of databases with NOW and the querying of such databases has been covered extensively elsewhere [10] However, the impact on modifications of the presence of NOW in the database as well as in the modification statements themselves has not been covered, even though many temporal data models, e.g. [3, 6, 7, 8, 17, 20, 26, 32], assume that NOW can be stored in the database. Before defining the semantics of modifications on variable databases, we specify three requirements to the accommodation of NOW . Requirement R1 The conventional insert, delete, and update statements should be extended to permit constant ....

M. H. Bohlen and C. S. Jensen. Seamless Integration of Time into SQL. Technical Report R-96--

....support of the DBMS may co exist harmoniously with the legacy code. Finally, a requirement aims at ensuring that programmers familiar with SQL 92 may start using the new features of the DBMS without a need for expensive training. Based on earlier work on a temporally extended SQL, termed ATSQL [4], we apply a set of more generic requirements to extend SQL 92 [17] to a spatio temporal SQL, termed STSQL. STSQL supports the two generic temporal aspects, valid time and transaction time, of database facts that record when facts are true in the modeled reality and when they are current in the ....

....ParaSQL [6] may be the closest relative of STSQL. Being based on an attribute value stamped data model, ParaSQL differs substantially from STSQL; apart from upward compatibility, it is our contention that it does not satisfy any of the migration requirements. Within temporal databases, ATSQL [4] and proposed additions to the SQL Temporal part of the SQL3 standard [24, 25] support bitemporal tables and also satisfy the migration requirements. STSQL may be seen as a generalization of these languages, its closest temporal relatives. Considering spatial data models, we have found no data ....

M. H. Bohlen and C. S. Jensen. Seamless Integration of Time into SQL. Technical Report R-96-49, Department of Computer Science, Aalborg University, Dec. 1996.

....attributes, and multiple space and time dimensions are permitted in a single table. The migration requirements that dictate the general properties of STSQL were originally developed in the context of bitemporal tables, i.e. tables supporting, at most, one transaction time and one valid time [2, 3]. STSQL supports multiple valid and transaction time and multiple space attributes in a single dimensional table. We are aware of no other models with this property. Among the few spatio temporal data models that exist, ParaSQL [5] may be the closest relative of STSQL. However, being based on an ....

....data models that exist, ParaSQL [5] may be the closest relative of STSQL. However, being based on an attribute value stamped data model, ParaSQL differs substantially from STSQL; apart from upward compatibility, it does not satisfy any of the migration requirements. STSQL generalizes ATSQL [3] and proposed additions to the SQL Temporal part of the SQL3 standard [22, 23] which support bitemporal tables and satisfy temporal migration requirements. Considering spatial data models, we have found no data models that provide migration support beyond upward compatibility. The ....

[Article contains additional citation context not shown here]

Bohlen, M. H., and Jensen, C. S. 1996. Seamless Integration of Time into SQL. Technical Report R-96-49, Department of Computer Science, Aalborg University.

No context found.

Bohlen, M. H., Jensen, C. S. Seamless Integration of Time into SQL. University of Aalborg, http://www.cs.auc.dk/ boehlen/Software/Tiger/atsql.ps.gz, 1996.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC