Re: A searchable datastructure for represeting attributes?

>> ... I know some systems working fine which are based on this
"mistake". That about problems you are speaking about - there are many other ways to verify data integrity and validity, up to application servers. So, the idea is
not that bad. Of course it is not for every system. <<

The other ways to validate data are not at the database level, so people can get around them. Even worse, the validation rules that Mr. A uses are not guaranteed to be the same validation rules that Mr. B uses.

I suppose if you had a small limited set of users, with a physically isolated machine, never changed the application and did not care about performance, or how long it takes to write code, you could keep this thing runing for more than 2 years.

Many years ago in Rapid City, South Dakota, Bjarne Stroustrup spoke to the INCITS H2 (nee ANSI X3H2) Database Standards Committee about C++. At the end of his talk, we asked about OO and databases. He said that Bell Laboratories, the finest minds in the world, had tried to create databases for objects and had four approaches. They all sucked. He then went into a speech about how OO was great for programming and the worst way to handle data. Received on Thu Feb 28 2002 - 23:12:11 CET