Re: [boost] [rfc] Quadrature library for numerical integrationofdefinite integrals

From: Hugo Duncan (hugoduncan_at_[hidden])
Date: 2007-12-07 10:31:10

On Fri, 07 Dec 2007 Paul A Bristow wrote:

> The effect on compile time is very bad (to me entirely unacceptable) and  
> the method of chaining (every member function returns
> *this) so you can write
> plot.color(red).title("my_title"). ...
>
> that Jake also used has proved much faster and just as convenient (in  
> fact it allows a much nicer and more flexible layout).
>
> So I strongly recommend against using Boost.Parameter based on this  
> experience.

Heeded.  Chaining could also be used to provide information on  
discontinuities, etc.

The current functional approach does however switch between  
implementations at compile time depending on whether an acceleration  
algorithm is supplied as a parameter or not.  I'll have to think about how  
to handle this in the chained approach ....