Re: Temporal extensions

Hi Jim,

On 27/04/15 21:48, Jim Nasby wrote:
> On 4/25/15 7:49 PM, Dave Jones wrote:
>> I've been working on a conversion of several utilities I wrote for
>> another engine, and was wondering if there was any interest in seeing
>> any of them added to contrib/ at some point in the vague undefined
>> future?
>
> Not in contrib, no, because there's no reason for these to be tied to
> specific versions of Postgres. Adding to PGXN would make sense though.

Thanks for the reply and suggestion - I've just run across PGXN having
finished reading through the all the temporal-related e-mails I could
find on pgsql-hackers for the last 5 years or so (specifically, Jeff
Davis' temporal extension). PGXN looks like a good place to park this
(once I've sorted out a few of the things I've run across while reading
- TRUNCATE triggers for it for starters!).

> (Though, I dislike using timestamps to do change/history tracking, but
> that's just me...)

I've been playing around with history tracking (in the context of BI,
typically with batch loaded reporting databases) for about 7-8 years now
and always found timestamps perfect for the purpose, but are you perhaps
referring to using it for audit purposes? If that's the case I'd agree
entirely - this is absolutely the wrong tool for such things (which is
something I need to put a bit more prominently in the docs - it's buried
in the design section at the moment).

Or did you mean ranges would be better? They certainly looked intriguing
when I started moving this stuff to postgres, and I'd like to re-visit
them in the near future as they offer capabilities I don't have with
timestamps (such as guaranteeing no overlapping ranges via exclusion
constraints) but my initial tests suggested some rather major
performance degradation so I put it on the back-burner at first.

Thanks,

Dave.