RE: [boost] Re: N-Dimensional array library submission...

From: Gary Powell (powellg_at_[hidden])
Date: 2002-01-08 18:51:27

ah!, sorry that I missed that. Multi-array seems to satisfy all of the
issues I can think of for now. I vote for inclusion.
  -gary-

PS
  Anyone up for a rerun of the tests with the code below? I'd be interested
if they are comparable in time, vs [].

-----Original Message-----
From: Jeremy Siek [mailto:jsiek_at_[hidden]]
Sent: Tuesday, January 08, 2002 4:04 AM
To: boost_at_[hidden]
Subject: RE: [boost] Re: N-Dimensional array library submission...

multi_array can do what you want, it has iterators over each dimension. It
looks like this:

for (i = A.begin(); i != A.end(); ++i)
  for (j = i->begin(); j != i->end(); ++j)
    for (k = j->begin(); k != j->end(); ++k)
      *k = values++;

On Mon, 7 Jan 2002, Gary Powell wrote:

> My problem is that for "C" code I tend to write:
>
> -------------------------------------------
>
> int carray_3D(double* A)
> {
>    // Assign values to the elements
>    int values = 0;
>    double const* const rowLength = (sizeof(double *) * row_size);
>    double const* const rowTimesCol = (rowLength  * col_size);
>    double const * const AEnd = A + (rowTimesCol * width_size));
>    for(double * i = A; i != AEnd; )
>    {
>      double const * const rowEnd = i + rowTimesCol;
>      for(double *j = 0; j != rowEnd; )
>      {
>        double *columnEnd = j + rowLength;
>        for(double *k = j; k != columnEnd; ++k)
>        {
>          *k = values++;
>        }
>        j = columnEnd;
>      }
>      i = rowEnd;
>    }
>
>    ...// etc.
>    return 0;
> }
>
> And that's what I want my multi-array access to generate. To not reload
> initial offsets, to do basic addition at the inner most level, not
> multiplication and addition for every access. (Without errors, as I'm sure
> the above code has!)
>
> IF you are only randomly accessing [][][], then I really like the syntax
of
> multi-array. It's just what I've written (Only not anywhere as nicely)
Call
> me greedy, but I want it all. Speed, easy to read source, and small code.
> (sounds like I need a better compiler!)
>
>   Yours,
>  -gary-
>
>
> Info: http://www.boost.org  Send unsubscribe requests to:
<mailto:boost-unsubscribe_at_[hidden]>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>

----------------------------------------------------------------------
 Jeremy Siek                          http://www.osl.iu.edu/~jsiek/
 Ph.D. Student, Indiana Univ. B'ton   email: jsiek_at_[hidden]
 C++ Booster (http://www.boost.org)   office phone: [phone]
----------------------------------------------------------------------

Info: http://www.boost.org  Send unsubscribe requests to:
<mailto:boost-unsubscribe_at_[hidden]>

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/