Re: identity and identifiers



Date view Thread view Subject view Author view

Perdita Stevens (Perdita.Stevens@dcs.ed.ac.uk)
Mon, 28 Feb 2000 13:41:18 +0000 (GMT)


I agree that this is an important distinction. Objects can have different identities without there being (at any given moment, in a natural sense) an identifier that distinguishes them; but if there is an identifier that distinguishes objects, then they must have different identities. I guess for our purposes it's probably reasonable to regard "having different identities" as interchangeable with the _conceivable possibility_ of there being an identifier that distinguishes. (In fact you could cook up a notion of identifier, in which all conceivable identifiers are held always to exist, that would make being indistinguishable by every identifier the same as having the same identity, but I think that would be cheating!) But in the context of specifying a sorting function -- where we came in -- I don't think the concept of identifier is helpful. We have no choice but to regard two objects as different if they have different identities, for a more basic reason. The reason is that we (the specifiers of the sorting function) can't legislate for all the identifiers in the system. As far as the sorting function is concerned, there may be no identifier that distinguishes two objects with the same state that occur in a bag to be sorted. But the caller of the sorting function, or some other part of the system, may indeed have an identifier that distinguishes them. I guess if we knew for sure that this wasn't the case in some precise sense, then we could swap them with impunity; but I don't think that fact has any practical relevance. Perdita


Date view Thread view Subject view Author view