Re: Unique handle generation

[Martin Oldfield <m@mail.tc>]

>>>>> "George" == George Belotsky <george@register.com> writes:

    George> Someone on this list talked about the possibility of
    George> people changing email addresses, and their old address
    George> being claimed by someone else.

    George> At first glance, an email address appears unique, but the
    George> above arguments do raise a legitimate concern.

Even if the old address isn't claimed I think that people changing
email addresses makes the concept flawed.

If someone changes their email address then either the handle changes
which is a pain, or it doesn't and the email address == handle
relation no longer holds. The latter case seems fraught with loads of
address qua handle and address qua address confusions.

Surely it's better to have a handle which really is immutable and tags
the object without any regard to the object's internal state ? To me
such a notion seems a thoroughly sensible concept for almost any
database. By contrast if the handle starts to relate to the object's
state then you probably end up reinventing an immutable tag for the
object too.

Cheers,
-- 
Martin Oldfield,
AdamsNames Ltd.