[Home] [By Thread] [By Date] [Recent Entries]
Michael, >> I'm agreed about effect on existing application - but i'm >> implying all time, that we are making comparison of XML and >> relational situation (instead of discussing of only one of >> them). If we move to XML databases, effect on existing >> application remains (non-triviality is not disappear !). >> MK> XML is far more flexible in this regard. MK> Well, if an <employee> has <nationality> as a child, then it can always have MK> two <nationality> children by a minor change to the schema. MK> This has far less impact on applications ... create table Employee ( id int primary key, family varchar ); create table Nationality ( id int primary key, nation varchar ); create table E2N ( e int references Employee(id), n int references Nationality(id) ); select Employee.Nationality from Employee.E2N.Nationality; OUTPUT: <employee id= family= > <nationality id= nation= > <nationality id= nation= > </employee> MK> ... than creating a separate table. Changing of XML-database <!attlist Employee Nationality> to <!element Employee (Nationality*) > (or even removal DTD at all) is less :) than to add two tables, but not in principal. And changing of request on SQL+XTree and on XQuery for new scheme is quite identical (no any language require much job). P.S. Maybe it's necessary to append auto-alter-table into R-DBMS and append DTD to control auto-alter-table to remove this little divergence in quantity of job ?? >> MK> It's also much easier to hold properties-of-properties MK> <employee> MK> <nationality date-acquired="2001-05-06" checked="no">British</nationality> MK> Adding attributes like this will very rarely affect existing applications. Like for the 'select * ...'; Dmitry Turin HTML6 (6. 5.4) http://html60.euro.ru SQL5 (5.11.1) http://sql50.euro.ru Unicode7 (7. 2.1) http://unicode70.euro.ru Computer2 (2. 0.2) http://computer20.euro.ru
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] |

Cart



