In article <[EMAIL PROTECTED]> [EMAIL PROTECTED] writes: Because if we really want sequentialization of the accesses and reuse of storage, we can (soon) use array transformers (cps), linear typesystems or perhaps array monads to express this at the source level. Will this be true if you have more than one array to update at a time? The first time I wrote Crout reduction I followed the definition given in a numerical analysis book which used two separate arrays to hold the _l_ and _u_ sections. After drawing little diagrams showing dependencies I realised they could be merged. However, there are other algorithms were it _seems_ that you need to be able to update more than one array at a time. Isn't the type signature you give incorrect? The parammeters m and n must have type a (where (a,a) is the index types of the arrays). I think your right, but as I can't seem to get to grips with the numeric hierarchy (anybody have a more detailed explanation of it?) I write with no signatures and see what the compiler produces. In this case Hbc 9.998.1 produced the signature I gave (Actually it has Integer instead of Int, but it was pointed out to me that Int is sufficient for indexing arrays) bevan
