tag:blogger.com,1999:blog-6411920579549337139.post5547637091363138119..comments2023-12-31T05:26:17.608-08:00Comments on DATABASE DEBUNKINGS: What Is a True Relational System (and What It Is Not)Fabian Pascalhttp://www.blogger.com/profile/01346669716885494092noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-6411920579549337139.post-25668854346032600892017-03-16T05:29:36.797-07:002017-03-16T05:29:36.797-07:00Regarding "the core being system-guaranteed l...Regarding "the core being system-guaranteed logical and semantic correctness of query and update results", it is easy to show that any dbms that defines an "insert" operation on a base relation as set union of itself and a set of tuples when another base relation has an existing tuple that does not agree on common attributes before the operation but does agree after the operation is not relational. Logically, such a result involves "deletion" aka set difference on the latter relation as well as the set union on the former. <br /><br />I imagine this logical error must include pretty much all dbms'es today, including ones that claim to be "truly" relational. (I think you have previously stated base updates should be avoided.) <br /><br />I call such systems pseudo-relational file systems. One of the characteristics of a file system is that any structurally compatible (in the physical sense) result is allowed, in other words what is not logically possible can be ignored, as in the above example where User semantics are allowed to trump logical results.toledobytheseahttps://www.blogger.com/profile/01660213928450837847noreply@blogger.com