Re: database design method

D Guntermann wrote:
>I see many advantages of value-based object identifiers, perhaps in such
>areas as identification (obvious), intrinsic ordering/mapping, as well as
>copies (deep versus shallow) of values and redundancy detection mechanisms.
>However, how would this be useful in a truly platform-independent,
>distributed datastore system? Is there a global mechanism for both
>generation and reconcilation of object identifiers - even if they avoid
>pointers and are based on value?

True object identifiers are abstract, i.e., the user never gets to see them. So you don't need a global mechanism and there is no reconciliation problem. You have to think of them as for example the nodes in a graph. A directed graph is a binary relation but how the nodes are exactly represented is immaterial.

• Jan Hidders