Re: I am confused as to whether I should use numeric IDs or not

Going back to the problem of IDs, part of the problem is to replicate whatever any individual has for 'recognition'.

If I ask you to think of 'Fred', you call up a cluster of ideas which recognise Fred.

It may be image recognition ,since you can recognise Fred from front or back, or pictorial recognition, since you can recognise 'his' handwriting, audial recogition, since you can recognise 'his' voice even over the phone or even 'concept' recognition , since you can have an idea about people you have never met. The point is that that bundle of recognition factors is attached uniquely to Fred.

The trouble is, do you assign a unique name to the bundle, or the bundle to a unique name.There is a difference. In one you recognise there is a bundle, in the other you recognise the bundle by its contents.

In many cases, (data transfer) all you want to do is recognise the bundle, in other cases ( data creation) all you need to know is that the bundle has a particular attribute or method, while in yet more cases ( data modification) you need to open the bundle, examine and modify its contents.

The difference now lies between the external Bundle label and the internal bundle label. Providing you have the method for recognising and tracing the bundle, it doesn't matter if the external bundle label changes. Your bundle 'Fred' becomes Frederick P. Smith when passed to another system and Passenger 57 when handed to a third.

On the other hand, you can get quite worried if the internal label changes and Bag, Passenger 57 ends up at Not Fred Destination.

Unfortunately it can be difficult to separate the external label and the internal label, or to persuade other people to do so. I had a problem where a plan went through several stages and changed its identity at each stage . Concept 1964 became Outline 763 became Project 99/281.3. The 'constant' was the fact that the papers all stayed in one file. It was a major problem to get the users to permit the idea of the bundle as an identity, since they saw the identity as changing. It then took a committee to identify a permissible external bundle label identification scheme even though it was completely irrelevant to any data they saw.

The task, of course, was to track the passage from Concept to Completion.

So some of the problem is taxonomic. We use the same words to describe the external and internal bundle labels and we 'need' a concrete description of the abtract 'another bundle'. Therein lies the opportunity for confusion.

Have fun
Robin

On Fri, 4 Feb 2000 18:51:57 +1100, "peter walker" <papwalker_at_ozemail.com.au> wrote:

>
>Chuck Grimsby <cdotgrimsby_at_worldnetdotattdot.net> wrote in message
>news:nauj9s4bau5td16uu3sm12othjgnajugoo_at_4ax.com...
>>
>> Errr.. Ahh... Peter.... I don't know how to tell you this, but...
>> ahh... Start worrying!
>Yeah, I can understand transactions having a unique id, I use them myself, I
>was referring to my account as an entity, not the transactions against the
>account. I can imagine my account number travels along through each wrapping
>(like ip in tcp/ip) in cross bank transactions. I was referring to the
>practice of giving us a unique visa number as the key to an account, not a
>combination of name on card, expirydate, etc to form the key. (Which may
>form another key as well..)
>
>I worry about banks anyway, the level of charges is inversely proportional
>to service, and service is in a geometric progression downwards here in Oz.
>
>An Australian bank is the closest thing to a perpetual motion machine I will
>ever see.
>
>peter walker
>
>
Received on Mon Feb 07 2000 - 00:00:00 CET