Re: [boost] "Software Development using the C++ Boost Library", book in preparatiion

Subject: Re: [boost] "Software Development using the C++ Boost Library", book in preparatiion
From: Mathias Gaunard (mathias.gaunard_at_[hidden])
Date: 2010-02-16 07:18:33

Daniel J. Duffy wrote:
> Mathias,'
> Thanks for the feedback.

I just realized my email might have sounded a bit harsh, sorry about 
this, wasn't my intention.

> We have Function to be very useful for PDEs; a PDE is modelled as a colection of Function instances which have different implementations dependent on the type of PDE (e.g. hear equation, convection, Black Scholes). Compared to function pointers and Bridge pattern this approach is much easier. And for semi-discretisation, a pde coefficient like f(x,y,z,t) transforms to F(t) when the special mesh is used (using Bind).

So it's for unified storage, and not the semi-discretisation itself. I'm 
just curious whether you could avoid the type erasure.

> How general are intrusive containers? Not just for embedded?

They're especially useful for system-level programming because you deal 
with the allocation and storage of the objects yourself, but of course 
their application isn't restrained to systems.