Re: [boost] Accelerating algorithms with SIMD - Segmented iterators and alternatives

Subject: Re: [boost] Accelerating algorithms with SIMD - Segmented iterators and alternatives
From: Mathias Gaunard (mathias.gaunard_at_[hidden])
Date: 2010-10-13 06:20:25

On 13/10/10 00:15, David Sankel wrote:

> Please please please, if you're going to attempt a better alternative for
> iterators, read "The essence of the Iterator pattern" [1] (or [2] for an
> updated version) and the important work on idioms[3] that it references. I
> wish I had more time to spend on this effort, but needless to say the tough
> semantic work has already been done!
>
> Thanks David A. for pointing out there was a semantics discussion going on
> here.
>
> David
>
> [1] http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.103.7480
> [2] http://www.comlab.ox.ac.uk/jeremy.gibbons/publications/iterator.pdf
> [3] http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.100.22
>

This is interesting in essence, but I fear it abstracts too much into 
lambda land. (to the point where I don't understand it, but that's 
another problem ;))

 From a glimpse at your paper, the advantages of your approach over a 
monadic map (which is what I take to be the most similar thing to a fold 
in C++, since C++ has side effects) is that it allows composition, and 
doesn't need two symmetric definitions to traverse the data both ways.
Is that correct?

Unfortunately I do no understand how one would use this in C++. The 
approach seems a bit similar to continuations, unless I'm completely out 
of it?
I have already suggested using coroutines to make the iteration more 
flexible, but it comes with a runtime cost (context switching).