From: gavino
Subject: mapreduce did lispers ever do this?
Date: Thu, 30 Nov 2006 08:26:37 +0000
Message-ID: <1164875197.130302.267000@n67g2000cwd.googlegroups.com>

http://labs.google.com/papers/mapreduce.html

• Re: mapreduce did lispers ever do this? Tim Bradshaw
• Re: mapreduce did lispers ever do this? John Thingstad
  • Re: mapreduce did lispers ever do this? Tim Bradshaw

From: Tim Bradshaw
Subject: Re: mapreduce did lispers ever do this?
Date: Thu, 30 Nov 2006 15:41:19 +0000
Message-ID: <1164901279.619707.42760@80g2000cwy.googlegroups.com>

gavino wrote:
> http://labs.google.com/papers/mapreduce.html

I've not read that paper recently, but I'm fairly sure that the
Connection Machines (at least the CM-1 and CM-2, later ones were rather
different and more conventional) supported this kind of operation, so
*Lisp would be worth looking at.  There was a simulator for it
somewhere, but I have no idea if it still exists.

--tim

From: John Thingstad
Subject: Re: mapreduce did lispers ever do this?
Date: Thu, 30 Nov 2006 17:52:17 +0000
Message-ID: <op.tjubxfbkpqzri1@pandora.upc.no>

On Thu, 30 Nov 2006 09:26:37 +0100, gavino <········@yahoo.com> wrote:

> http://labs.google.com/papers/mapreduce.html
>

You might note that the tecnical director of Google is Peder Norvig.  
(PAIP, AIMA, www.norvig.com)
The map-reduce algorithm is written in C for speed.
There is little doubt that a few lessons were learn't from Lisp however.

-- 
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

From: Tim Bradshaw
Subject: Re: mapreduce did lispers ever do this?
Date: Thu, 30 Nov 2006 19:53:37 +0000
Message-ID: <ekncs3$dpu$1$8300dec7@news.demon.co.uk>

On 2006-11-30 17:52:17 +0000, "John Thingstad" <··············@chello.no> said:

> The map-reduce algorithm is written in C for speed.

I doubt for speed.  This is something that runs across huge clusters, 
performance will be dominated by I/O.  There are plenty of other 
reasons to write it in C but I doubt speed was one.