On 08/24/2015 08:44 PM, Rob Sargent wrote:
On Aug 24, 2015, at 6:53 PM, Melvin Davidson <melvin6925@xxxxxxxxx> wrote:
You are right, he was probably talking about FK's. I was just so frustrated about people insisting that using "ID" as the primary key in every table is a "good" idea,
I didn't bother to reply previously. I stand firm on my belief that the primary key should be something meaningful and NOT "id" just for the sake of having a unique numeric key.
What, pray tell, is the unique natural key of person in any meaningfully large domain such as state? Certainly not name + birthdate. Current address isn’t guaranteed. Social isn’t reliable and actually not truly unique.
To add:
1) Who determined that a number is not natural?
2) One of the older unique natural keys (genus, species) is not so
unique. I am a fisheries biologist by training and in my time the
'unique' identifier for various fishes has changed. Now that
ichthyologists have discovered DNA testing, it can be expected there
will be even more changes. This is even more apparent when you go back
in in history. As an example:
https://en.wikipedia.org/wiki/Rainbow_trout
Rainbow trout
Current
Oncorhynchus mykiss
Past
Salmo mykiss Walbaum, 1792
Parasalmo mykiss (Walbaum, 1792)
Salmo purpuratus Pallas, 1814
Salmo penshinensis Pallas, 1814
Parasalmo penshinensis (Pallas, 1814)
Salmo gairdnerii Richardson, 1836 <--The one I learned.
Fario gairdneri (Richardson, 1836)
Oncorhynchus gairdnerii (Richardson, 1836)
Salmo gairdnerii gairdnerii Richardson, 1836
Salmo rivularis Ayres, 1855
Salmo iridea Gibbons, 1855
Salmo gairdnerii irideus Gibbons, 1855
Salmo irideus Gibbons, 1855
Trutta iridea (Gibbons, 1855)
Salmo truncatus Suckley, 1859
Salmo masoni Suckley, 1860
Oncorhynchus kamloops Jordan, 1892
Salmo kamloops (Jordan, 1892)
Salmo rivularis kamloops (Jordan, 1892)
Salmo gairdneri shasta Jordan, 1894
Salmo gilberti Jordan, 1894
Salmo nelsoni Evermann, 1908
All the above point to the same fish and have appeared and appear in
articles and reports about said fish. Lets not even get into the common
name situation:).
Even given that there are models which are made of entities with legitimate attributes which per force define a unique instance, I see no benefit in avoiding the convenience of an arbitrary and simple value for the key. Is it the overhead of generating and storing one more value per tuple that you can’t abide?
--
Adrian Klaver
adrian.klaver@xxxxxxxxxxx
--
Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general
