From: Neil Conway <nconway@...>
Date: 2001-12-09T14:35:22+09:00
Subject: [ruby-talk:27998] Re: DBI and large result sets

On Sat, 2001-12-08 at 22:45, Albert Wagner wrote:
> On Saturday 08 December 2001 09:28 pm, you wrote:
> > On Fri, 2001-12-07 at 20:40, Ned Konz wrote:
> > > (note: MySQL is built on top of BerkeleyDB).
> >
> > This is OT, but that's quite true. MySQL can create tables using one of
> > several different table types. _One_ of these uses BerkeleyDB to support
> > transactions, but if you're not using transactions it's suggested that
> > you avoid it. And even if you want transactions, there may be other
> > tables types that can do transactions without involving BerkeleyDB
> > (maybe Gemini? I'm a PostgreSQL user, I'm only vaguely familiar with
> > MySQL).
> 
> Why is it suggested that you avoid BerkeleyDB?

IIRC, because you pay a speed, memory and disk space penalty over the
default MySQL table type (MyISAM). In exchange, you get transactions,
but if you don't need those (and MySQL users typically don't), then you
may be better of without BDB tables.

And last I checked, the code for BDB tables wasn't tested as thoroughly
as the code for MyISAM, but that may well have changed.

(And now, back to your regularly scheduled programming ...)

Cheers,

Neil

-- 
Neil Conway <neilconway@rogers.com>
PGP Key ID: DB3C29FC