Re: [boost] Re: loki::type_list == boost::mpl::list_node

From: David Abrahams (david.abrahams_at_[hidden])
Date: 2002-04-16 10:43:17

----- Original Message -----
From: "Hamish Mackenzie" <hamish_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Tuesday, April 16, 2002 9:21 AM
Subject: Re: [boost] Re: loki::type_list == boost::mpl::list_node

> On Tue, 2002-04-16 at 14:45, David Abrahams wrote:
> >
> > It doesn't surprise me too much for such a short list. These facts
> > contribute:
>
> Can these explain the fact that longer lists are even slower?  I've
been
> out playing tennis and couldn't stop thinking there must be a bug
> somewhere.  Is my code ok?
>
> N=1 4.83
> N=2 4.97
> N=3 5.26
> N=4 12.10
> N=5 12.83
> N=6 13.76
> N=7 13.86
> N=8 101.34
> N=9 112.61
>
> Something ain't right this should be O( N ) or O( abs( N/4 ) ).  Looks
> more like O( abs( N/4 ) ^ 3 ).

It doesn't surprise me. See
http://users.rcn.com/abrahams/instantiation_speed/index.html. The
implementations of template matching in some of these compilers is...
well, let's just say it wasn't designed with metaprogramming in mind.
Lots of linear list searches, for example.

-Dave