ch-ch-update: series, closed-form series, and strides

[Bill Baxter]

On Sun, Feb 1, 2009 at 3:30 AM, Don <nospam at nospam.com> wrote:
> Andrei Alexandrescu wrote:
>>
>> Jagged, banded, diagonal, fixed-size, and sparse layouts come naturally
>> and will be hopefully supported in a mixable-and-matchable form (e.g. I want
>> a block matrix of 3x3 matrices). I want to focus on storage for now, put it
>> in the standard library, to then allow great scientific programmers like you
>> guys to use those storages as a common data format on top of which cool
>> algorithms can be implemented.
>
> This sounds like the perfect approach. Everything's built around the storage
> format.

Yes indeed.  It's a tried and true approach.  I think you'll find a
similar approach in the likes of the Matrix Template Library, or
FLENS.  But I think those projects, particularly MTL, hit a wall
because C++ meta-programming is just such a mess.  I'm very excited to
see what can be done with the full arsenal of D2 metaprogramming at
one's disposal.

--bb