| From: | Greg Stark <gsstark(at)mit(dot)edu> |
|---|---|
| To: | Leandro Guimarães Faria Corcete DUTRA <leandro(at)dutra(dot)fastmail(dot)fm> |
| Cc: | pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: Surrogate keys (Was: enums) |
| Date: | 2006-01-15 05:08:42 |
| Message-ID: | 87lkxixelh.fsf@stark.xeocode.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Leandro Guimarães Faria Corcete DUTRA <leandro(at)dutra(dot)fastmail(dot)fm> writes:
> Certainly decoupling presentation from storage would be nice, but even before
> that generalised use of surrogate keys seems to me a knee-jerk reaction.
I hate knee-jerk reactions too, but just think of all the pain of people
dealing with databases where they used Social Security numbers for primary
keys. I would never use an attribute that represents some real-world datum as
a primary key any more.
In my experience there are very few occasions where I want a real non-sequence
generated primary key. I've never regretted having a sequence generated
primary key, and I've certainly had occasions to regret not having one.
--
greg
| From | Date | Subject | |
|---|---|---|---|
| Next Message | korry | 2006-01-15 16:14:05 | Coding standards? Recommendations? |
| Previous Message | Qingqing Zhou | 2006-01-14 21:13:56 | Re: Warm-up cache may have its virtue |