[Info-ingres] Re: is there an equivavlent to auto_increment in ingres ?
- From: Michael Leo <mleo@xxxxxxxxxxxxxxx>
- Date: Wed, 20 Jul 2005 07:27:46 -0500
At 2:30 AM -0700 7/20/05, --CELKO-- wrote:
>> A better example is FedEx/UPS/DHL. A package's "tracking number" is a definite surrogate key. No doubt about it. <<
No, no, no; those are not surrogate keys. Go back to my post and read Dr. Codd's definition. They are "intelligent keys" that can be put in a bar code and used by machinery to track and to route a package. They have check digits, the destination, etc. in the created key.
In the book trade, we have the SAN (Standard Address Number) which was used for shipping and controlled by a trusted outside source. Bookdealers, publishers, etc. were not in the shipping, so this was pretty handy. FedEx/UPS/DHL et al **are** the shipping business, so they have to be their own internal trusted source.
I'm not sure I agree. I don't think FedEx/UPS/DHL considers the "tracking number" as merely an intelligent key used by machinery.
Some experience with their systems tells me they are used as the primary key for everything.
Remember the consumer experience with shipping is very simple. 99% of us never have more than one item in transit at any one time.
But shippers work with multi-national companies that ship 100,000 items per day from 100 locations to 30,000 destinations at varying urgencies, transports, and shipping specifications.
Thinking about it, I guess they aren't surrogate keys, as they are exposed to the user and not sitting side by side with the logical primary key. I suspect they started as surrogate keys.
When the truly logical primary key is so hideously verbose an complex as a shipped package's tracking identifier, the human mind (in my opinion) reverts to something easier to represent it. I don't think anyone relates to package "12395757631923" by anything other than that number. I don't think people mentally substitute the dozen or so unique fields that logically distinguish one shipped package from another.
So I'm not sure if we agree or disagree. But I really can't see how we can live without abstract "keys" exposed to users in everyday life.
Cheers,
-- Michael Leo Java, J2EE, BEA WebLogic, Caribou Lake LLC Oracle, Open Source, Ingres, mleo@xxxxxxxxxxxxxxx Real Enterprise Applications .
- References:
- is there an equivavlent to auto_increment in ingres ?
- From: morgan brickley
- Re: is there an equivavlent to auto_increment in ingres ?
- From: --CELKO--
- Re: is there an equivavlent to auto_increment in ingres ?
- From: Mike Lay
- [Info-ingres] Re: is there an equivavlent to auto_increment in ingres ?
- From: Michael Leo
- Re: is there an equivavlent to auto_increment in ingres ?
- From: --CELKO--
- is there an equivavlent to auto_increment in ingres ?
- Prev by Date: Re: [Info-ingres] Re: is there an equivavlent to auto_increment in ingres ?
- Next by Date: Re: [Info-ingres] Re: is there an equivavlent to auto_increment i n ingres ?
- Previous by thread: Re: is there an equivavlent to auto_increment in ingres ?
- Next by thread: RE: [Info-ingres] Re: is there an equivavlent to auto_increment i n ingres ?
- Index(es):
Relevant Pages
|