{"title":"Comments on the debate: data-structure-set versus relational","authors":"J. Clymer","doi":"10.1145/984403.984406","DOIUrl":null,"url":null,"abstract":"In the belief that it might be worthwhile for someone to attempt a summary of the \"Great Debate,\" I have tried to set down my own observations and wlll pass them on to you for whatever they may be worth. Suppose we consider that there wlll be three supported interfaces to future data base systems, and describe these suggestively as an End User Interface, an Application Programmer Interface, and a Systems Programmer Interface. Table 1 shows how these might very roughly be characterized. (Note: By systems programmer I am referring to the data base management system and not to the operating system.) There seems to be little argument that the End User Interface should have a basically relational view, or that some form of network view will be required for the Systems Programmer Interface. This is because the concept of independence from logical data structure more or less implies the relational view for the higher level~, and the requirement to eliminate redundancy from the stored data more or less implies a network view for the lower level. I gather, then that the central thrust of the debate is over what form the Application Programmer Interface should take. This does seem to be a substantive issue, and not merely a \"religious\" argument. Dr. Codd's position seems to be that the application programmer should be able to request data In relational terms and have it supplied by the data management facility into his work area, where he may then operate upon it in. whatever fashion he wishes; or that he be able to call upon the data management facility to take data from his work area described in relational terms and store it into the data base. To the extent that the data base contains linkages and indices required to reduce redundancy and improve efficiency, it will be the responsibility of the data management facility to manage these, and they should be essentially invisible to the application programmer. In this way the viewpoint of the application programmer and the end user will be essentially the same, and communication between them will be easy. The task of the application programmer will be reduced and he will make fewer mistakes because he will not have to be concerned with how he acquires the data he wants to manipulate, but only with its specification.","PeriodicalId":123835,"journal":{"name":"FDT Bull. ACM SIGFIDET SIGMOD","volume":"7 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1975-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"FDT Bull. ACM SIGFIDET SIGMOD","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/984403.984406","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
Abstract
In the belief that it might be worthwhile for someone to attempt a summary of the "Great Debate," I have tried to set down my own observations and wlll pass them on to you for whatever they may be worth. Suppose we consider that there wlll be three supported interfaces to future data base systems, and describe these suggestively as an End User Interface, an Application Programmer Interface, and a Systems Programmer Interface. Table 1 shows how these might very roughly be characterized. (Note: By systems programmer I am referring to the data base management system and not to the operating system.) There seems to be little argument that the End User Interface should have a basically relational view, or that some form of network view will be required for the Systems Programmer Interface. This is because the concept of independence from logical data structure more or less implies the relational view for the higher level~, and the requirement to eliminate redundancy from the stored data more or less implies a network view for the lower level. I gather, then that the central thrust of the debate is over what form the Application Programmer Interface should take. This does seem to be a substantive issue, and not merely a "religious" argument. Dr. Codd's position seems to be that the application programmer should be able to request data In relational terms and have it supplied by the data management facility into his work area, where he may then operate upon it in. whatever fashion he wishes; or that he be able to call upon the data management facility to take data from his work area described in relational terms and store it into the data base. To the extent that the data base contains linkages and indices required to reduce redundancy and improve efficiency, it will be the responsibility of the data management facility to manage these, and they should be essentially invisible to the application programmer. In this way the viewpoint of the application programmer and the end user will be essentially the same, and communication between them will be easy. The task of the application programmer will be reduced and he will make fewer mistakes because he will not have to be concerned with how he acquires the data he wants to manipulate, but only with its specification.