Re: range remove in BTree, or other table storage
Date: 23 Feb 2007 09:10:43 -0800
Message-ID: <1172250643.734351.101120_at_k78g2000cwa.googlegroups.com>
On Feb 23, 3:42 pm, Sampo Syreeni <d..._at_iki.fi> wrote:
> On 2007-02-23, Cimode wrote:
> > The most efficient way to do bulk inserts is interset fusion.
>
> What's that? I've probably stumbled onto it, but under a different name.
>
> > As B-trees are themselves hacks because they do not separate logical
> > and physical, the chances you will be able to build efficient bulk
> > insert through b-trees are NULL.
>
> I don't really understand this either. I've always thought of B-trees as
> a physical implementation detail, so I don't see where they contain
> anything logical at all. Especially if hidden behind some neat
> abstraction, like a GiST.
Any physical implementation (BTree) relies on some logical computing
model. Any computing model relies on a logical model of higher model
of abstraction. Gist is neither a computing model nor it has a
logical abstract background. What are the chances an implementation
issued from it may be anything else than a hack.
Received on Fri Feb 23 2007 - 18:10:43 CET