Re: overloading [] operator for matrix?

From: Cy Edmunds (cedmunds_at_spamless.rochester.rr.com)
Date: 02/28/04 

Date: Sat, 28 Feb 2004 05:45:28 GMT

"Dave" <davis_eric@yahoo.com> wrote in message
news:99f6e996.0402272108.60879104@posting.google.com...
> Hi there:
>
> i am doing a matrix class for practice. everything is ok except the []
> operator. Basically, i want to achieve the same way as the normal
> double[]'s, such as:
> m[2][3] = 4.0;
> m[3][4] = m[4][5] + m[2][3];
> cout<<m[2][3]<<endl;
> I also want the indices to be 1 based instead of 0 based.
>
> One way I can think out is to have the following overloaded operator
> in matrix class:
> double* operator [] (int row_index); suppose the matrix contains
> doubles and i am using a double pointer to store the numbers.
>
> but I have the following problems with this approach:
> 1. the second index can't be 1 based since it's used to directly
> indexed into the double*. It has to be 0 based.

Why? Can't you make the pointer point one location before the actual data?
We did it for years in Fortran. Of course there is some inefficiency with
computing the offset, but that's 1-based indexing for you.

> 2. it's not safe. since the function returns the row pointer. the
> caller can just call m[2] and then do some bad things.

Don't look now but indexing isn't safe. It's just pointer arithmetic in
disguise.

>
> but that's the only way I can think out now. Is there a good way to
> solve this problem?

You could have your operator [] return a "row" class representing a row in
your matrix. There might be some advantage to this for implementing
operations such as row swapping. However it's not going to be more efficient
for indexing than just returning a pointer and may be considerably less so.

>
> Thanks a lot. 

-- 
Cy
http://home.rochester.rr.com/cyhome/

Relevant Pages

  • Re: Programming Language Productivity: The Stupidity of Programmers
    ... So indexing retains more information. ... The assertion made was "indexing retains more information" than pointer ... Neither is more amenable to compiler optimization. ...
    (comp.programming)
  • Re: strtok and strtok_r
    ... Passing in a NULL should be a NOP in my view. ... you don't want to be indexing ... the NULL pointer (for one thing, ...
    (comp.lang.c)
  • Re: Fast Image access for binarization
    ... yes we did replace the indexing by a pointer. ... but it still wont give us ... something on the lines of sub-sampling? ...
    (sci.image.processing)
  • Re: More on pointers to pointers.
    ... Dave Vandervies wrote: ... it will be converted to a pointer to its first ... In the case of an array, the second version makes explicit what's being ... since array indexing is defined in terms of pointer addition: ...
    (comp.lang.c)
  • Le TREC - Result
    ... Got to the yard yesterday and was welcomed by a muddy bog monster, ... Bramble came out and when Dave went it, he started again, wehn Dave ... came out and we went in, poor Pointer eyes were on stalks. ... nowhere near it, so left it at that, exit fast! ...
    (uk.rec.equestrian)
Loading