Declarative specifications of complex transactions

While specifications of queries are usually of a declarative nature (since the work of Codd in the early seventies), specifications of transactions are still mainly of an operational and descriptive nature. Especially descriptions of complex transactions (such as cascading deletes) tend to be very o...

Full description

Bibliographic Details
Main Author: Brock, E.O. de
Format: Report
Language:unknown
Subjects:
DML
Online Access:http://irs.ub.rug.nl/ppn/174867816
id ftrepec:oai:RePEc:dgr:rugsom:98a41
record_format openpolar
spelling ftrepec:oai:RePEc:dgr:rugsom:98a41 2024-04-14T08:10:56+00:00 Declarative specifications of complex transactions Brock, E.O. de http://irs.ub.rug.nl/ppn/174867816 unknown http://irs.ub.rug.nl/ppn/174867816 preprint ftrepec 2024-03-19T10:33:40Z While specifications of queries are usually of a declarative nature (since the work of Codd in the early seventies), specifications of transactions are still mainly of an operational and descriptive nature. Especially descriptions of complex transactions (such as cascading deletes) tend to be very operational. Often these descriptions turn out to be nondeterministic as well. A problematic consequence is that the semantics of transactions and of several related notions are often unclear or even ambiguous. For a database designer this surely is not a good starting point for building applications. Another tendency we recognize is that the current literature on transactions is mainly driven by technical solutions offered by research prototypes and commercial systems and not so much by advanced specification requirements from a user's or database designer's point of view. In our opinion, the research questions should (also) include what kind of complex transactions (advanced) users would like to specify (and not only what e.g. the expressive power of a given technical solution is), and how these specifications can be translated to implementations in the currently available (advanced) databasemanagementsystems. And, moreover, wasn't it our purpose (with the introduction of 4GL's and the like) to become declarative instead of operational, concentrating on the "what" instead of the "how"? This paper offers a general framework for declarative specifications of transactions, including complex ones. We also take the influence of static and dynamic constraints on the alleged transactions into account. The paper introduces sharp (deterministic) definitions of (and distinctions between) compound transactions and a sequence of transactions. Basic set theory is our main vehicle. Applications of our theory include the provision of a formal semantics of SQL's DML-statements (which include restricted deletes), the declarative specification of cascading deletes, and the distinction between allowable and available transitions. ... Report DML RePEc (Research Papers in Economics)
institution Open Polar
collection RePEc (Research Papers in Economics)
op_collection_id ftrepec
language unknown
description While specifications of queries are usually of a declarative nature (since the work of Codd in the early seventies), specifications of transactions are still mainly of an operational and descriptive nature. Especially descriptions of complex transactions (such as cascading deletes) tend to be very operational. Often these descriptions turn out to be nondeterministic as well. A problematic consequence is that the semantics of transactions and of several related notions are often unclear or even ambiguous. For a database designer this surely is not a good starting point for building applications. Another tendency we recognize is that the current literature on transactions is mainly driven by technical solutions offered by research prototypes and commercial systems and not so much by advanced specification requirements from a user's or database designer's point of view. In our opinion, the research questions should (also) include what kind of complex transactions (advanced) users would like to specify (and not only what e.g. the expressive power of a given technical solution is), and how these specifications can be translated to implementations in the currently available (advanced) databasemanagementsystems. And, moreover, wasn't it our purpose (with the introduction of 4GL's and the like) to become declarative instead of operational, concentrating on the "what" instead of the "how"? This paper offers a general framework for declarative specifications of transactions, including complex ones. We also take the influence of static and dynamic constraints on the alleged transactions into account. The paper introduces sharp (deterministic) definitions of (and distinctions between) compound transactions and a sequence of transactions. Basic set theory is our main vehicle. Applications of our theory include the provision of a formal semantics of SQL's DML-statements (which include restricted deletes), the declarative specification of cascading deletes, and the distinction between allowable and available transitions. ...
format Report
author Brock, E.O. de
spellingShingle Brock, E.O. de
Declarative specifications of complex transactions
author_facet Brock, E.O. de
author_sort Brock, E.O. de
title Declarative specifications of complex transactions
title_short Declarative specifications of complex transactions
title_full Declarative specifications of complex transactions
title_fullStr Declarative specifications of complex transactions
title_full_unstemmed Declarative specifications of complex transactions
title_sort declarative specifications of complex transactions
url http://irs.ub.rug.nl/ppn/174867816
genre DML
genre_facet DML
op_relation http://irs.ub.rug.nl/ppn/174867816
_version_ 1796308599356522496