Re: a union is always a join!
Date: Wed, 18 Mar 2009 22:00:32 -0700 (PDT)
Message-ID: <557d6f32-dccf-4512-b74e-2e6e1cb064a0_at_w35g2000prg.googlegroups.com>
On Mar 18, 6:25 pm, br..._at_selzer-software.com wrote:
> I think that the delete/insert pair states that
> there is a different thing with key p = 1, whereas the update states
> that the thing with key p = 1 merely appears different. In other
> words, the delete/insert pair describes two distinct things, whereas
> the update describes two distinct appearances of the same thing. The
> delete/insert pair,
>
> delete {(p=1,q=a)}
> insert {(p=1,q=a)}
>
> states that there are two distinct things with exactly the same
> components, just during adjacent intervals.
> Unfortunately, it just isn't that simple. Since the key components
> that identify something can be different at different times yet still
> identify that same something, there can be more than one transition
> that yields a resulting state. For example, if the guy that had up to
> this point been first in line at the bank was wearing a blue hat, and
> if the guy that is now first in line is wearing a red hat, then one
> possibility is that there is a different guy that is first in line,
> the guy wearing the red hat, but another possibility is that the guy
> that had up to this point been first in line is still first in line
> but just put on a red hat. So which is it? If the transition
> consisted of a delete/insert pair, then it's clear that it's the first
> possibility, but if the transition consists of an update, then it's
> clearly the second.
This "real life" example is not helping. I doubt customer hat is of any concern to the bank business (unless it is ski hood), much less to the general public. Received on Thu Mar 19 2009 - 06:00:32 CET