Re: PostgreSQL Developer Best Practices

On 08/25/2015 02:23 PM, Gavin Flower wrote:
> On 26/08/15 08:56, Adrian Klaver wrote:
>> On 08/25/2015 08:40 AM, Melvin Davidson wrote:
>>> Adrian,
>>>
>>> Stop being so technical. When we/I speak of natural keys, we are talking
>>> about the column
>>> that would NATURALly lend itself as the primary key.
>>
>> Pretty sure this is a technical list:)
>>
> Don't let inconvenient facts get in the way of a good argument! :-)

>> have all gone to the same seminar on how to be Walmart and decided
>> they did not want unique numbers, but UPCs tied to price groups that
>> covered a variety of plants. Luckily, I was too stupid to
> Natural Stupidity??? :-)

Oh yeah and a long history too, but that needs at least a pitcher of
beer to recount.

>
> (Sorry, couldn't resist!)
>
>> know surrogate keys where bad and had a sequence attached to the tag
>> table. This then became the tag id and made life a lot easier during
>> the transition. It still remains there, because people are people and
>> 'natural' tends to be artificial and transient.
> Extremely good examples, I'll bear them in mind - makes me even more
> keen on surrogate primary keys. I'm always very wary when people tell
> me some numbering scheme will NEVER change!!!

To add a recent one. My partner Nancy signed up for Medicare last year
to avoid the premium penalty. This year in July she signed up for Social
Security. Turns out, for reasons I do not understand,
CMS(https://www.cms.gov/) changes the Medicare account number at that
point. The reason we even cared is that the billing system thinks she
has two accounts and is double billing. Time on the phone with someone
at CMS was not enlightening. We where told to trust the system and
eventually it will work itself out. Still waiting:(

>
> [...]
>
>
> Cheers,
> Gavin
>

--
Adrian Klaver
adrian(dot)klaver(at)aklaver(dot)com