Mailing-List: contact dev-help@couchdb.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@couchdb.apache.org
Received-SPF: pass (athena.apache.org: domain of b.candler@pobox.com
 designates 208.72.237.25 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws; d=pobox.com; h=date:from:to
	:subject:message-id:references:mime-version:content-type
	:in-reply-to; q=dns; s=sasl; b=XhrTknEH53RWccqmbYCSzxg0AKFFi4k72
	sTkZB2e5HvthQy3izn8b5o9onxW7ECVNrDwamVN/bJ2wleC6VYrNw3FxyoQjSECb
	fjqZdhXCoMoBZjZzCVWaTCr5JlHlFCjrOX2OmRhK2eVqIE2+0ST7PXmYwN+SqOJq
	7JWpYOsDC0=
Date: Thu, 4 Feb 2010 11:01:29 +0000
From: Brian Candler <B.Candler@pobox.com>
To: dev@couchdb.apache.org
Subject: Re: DB ACLs (was Re: 0.11 Release / Feature Freeze for 1.0)
Message-ID: <20100204110129.GA4844@uk.tiscali.com>
References: <e282921e1002030921g53af1711r3b46ebe31f4cb38b@mail.gmail.com>
 <20100203212426.GA10515@uk.tiscali.com>
 <A9CCB20B-A81E-47C9-91D8-0D0E40DAF253@apache.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <A9CCB20B-A81E-47C9-91D8-0D0E40DAF253@apache.org>
User-Agent: Mutt/1.5.20 (2009-06-14)

On Wed, Feb 03, 2010 at 02:27:56PM -0800, Jan Lehnardt wrote:
> > Sorry to be blunt, but do you have numbers to back that up?  This smells
> > very much of premature optimisation.
> 
> Reading a document costs: 
> 
>     1 disk seek to the end of the db file to grab the b-tree root
>  + n disk seeks to the document (n < 5 to 10 in most cases)
> 
> In a well cached database, most of the seeks are free, but the last one
> usually isn't (unless you can hold a full database in memory).

That's not true here. If _readers were a document within the database then
the *same* document would be read for every access, and therefore it would
be cached.  There's no seek.

But again you've ignored my main point: I'm proposing that the authorisation
be a role within the userctx, and the userctx is already loaded.  Free is
better than cheap.

Actually, that's not entirely true. We want to have publicly-readable
databases, so we need an "_anon" user with roles assigned to it, and the
effective roles you get would be the union of the roles from your user
record plus the roles from the _anon user.

That might mean reading the _anon record from the _users database.  That
will always be the same record, so is bound to be cached.

Regards,

Brian.