On Tue, Jun 17, 2014 at 2:16 PM, Andres Freund <and...@2ndquadrant.com> wrote:
>> But I'm not trying to say that we absolutely have to support that kind
>> of thing; what I am trying to say is that there should be a README or
>> a mailing list post or some such that says: "We thought about how
>> generic to make this. We considered A, B, and C. We rejected C as
>> too narrow, and A because if we made it that general it would have
>> greatly enlarged the disk footprint for the following reasons.
>> Therefore we selected B."
> Isn't 'simpler implementation' a valid reason that's already been
> discussed onlist? Obviously simpler implementation doesn't trump
> everything, but it's one valid reason...
> Note that I have zap to do with the design of this feature. I work for
> the same company as Alvaro, but that's pretty much it...
It really *hasn't* been discussed on-list. See these emails,
discussing the same ideas, from 8 months ago:
Now, Alvaro did not respond to those emails, nor did anyone involved
in the development of the feature. There may be an argument that
implementing that would be too complicated, but Heikki said he didn't
think it would be, and nobody's made a concrete argument as to why
he's wrong (and Heikki knows a lot about indexing).
>> Basically, I think Heikki asked a good
>> question - which was "could we abstract this more?" - and I can't
>> recall seeing a clear answer explaining why we could or couldn't and
>> what the trade-offs would be.
> 'could we abstract more' imo is a pretty bad design guideline. It's 'is
> there benefit in abstracting more'. Otherwise you end up with way to
> complicated systems.
On the flip side, if you don't abstract enough, you end up being able
to cover only a small set of the relevant use cases, or else you end
up with a bunch of poorly-coordinated tools to cover slightly
different use cases.
The Enterprise PostgreSQL Company
Sent via pgsql-hackers mailing list (email@example.com)
To make changes to your subscription: