Newbie question on multidimensional arrays

[Sudarshan Raghunathan]

Hello,

I'm trying to understand how subsref and subsasgn are implemented for
multidimensional arrays in Octave. In other words, what exactly is the
magic required to implement
B = A(m1:dm:m2, n1:dn:n2, ...) for ndims(A) >= 3? Of course, the
naive way would be to generate all n-tuples and then do a sub2ind, but
I suspect that would be too slow. I tried browsing through the source
code, but did not find where exactly the enumeration was happening.
Could someone please let me know where I could find more information?

Many thanks,
Sudarshan