|
|
Ublas : |
From: Jeffrey Brent McBeth (mcbeth_at_[hidden])
Date: 2005-06-22 05:43:44
On Wed, Jun 22, 2005 at 12:17:58PM +0200, Michael Stevens wrote:
> The problem now is that people are 'surprised' that either code is suboptimal
> with regard to speed and temporaries.
>
> I think it is hard to avoid surprises!
The current design avoids destructive surprises. I'll take "works every time"
over "subtley broken" any day. The fact that uBlas allows you to optimize
when you know things will be okay, and that this is well documented (you
can't miss it when investigating the library) is just icing on the cake.
Look, if I make a mistake with a matrix operation, I permanently destroy
someone's vision and probably livelyhood. We do rigorous tests to avoid
such an occurence, but anybody that believes that tests can cover 100% of
cases are deluded. The fact that by default, uBlas does the right thing, is
additional confidence in what we do.
Right first, speed second.
-- ---------------------------------------------------------------------------- Computer Science is as much about computers as astronomy is about telescopes -- Edsger Wybe Dijkstra (1930-2002) ----------------------------------------------------------------------------