Re: String versus integer joins?

I guess my point is that there is no simple answer - it's all weeds. :) That being said, I doubt the text vs integer thing matters much, especially your joins are indexed.

d.

On Oct 23, 2012, at 13:10, Veronika Megler <vmegler(at)cecs(dot)pdx(dot)edu> wrote:

> Actually, right now I'm trying to avoid the weeds, and all the other factors except this one (since I have other sources for those other factors).
>
> Back in the distant past, on a non-PostgreSQL high-performance database, there used to be very simple guidance: "don't use character-based fields as a unique tuple identifier, because performance will suck. Use integer identifiers."
>
> But the years have passed, and I keep hearing vague statements that things are much better now, and maybe it isn't true these days, and anyway, that wasn't PostgreSQL; but then maybe it might make a difference after all.
>
> What I was hoping for was some more definitive statement or someone's actual experience.
>
> If there's a statement on this specific topic in the High Performance section I'd be very interested to hear what it is,
> Veronika