Andrew U. Frank <[EMAIL PROTECTED]>
write on 24 Aug 2000 on basAlgPropos
> i am interested in an improvement of the numeric classes and hope the
> proposal by Mechveliani can contribute to this.
basAlgPropos contains such suggestion.
> [..]
> i would suggest that the authors separate the proposal into two
> parts - the piece which identifies the parts which must be changed
> in the prelude and a library of routines, which demonstrate that on
> the proposed base functionality in the prelude an algebraic package
> can be built.
> [..]
> It seems to me that all what is needed to change in the Prelude are
> the classes defining numbers (Num, Rationl, Fractional...);
> everything else can go in a support library? is this correct?
Yes, probably.
The proposal contains
(0) common purpose items (data PropValue, sortE, ...)
(1) algebraic classes (Set, ..., EuclideanRing, Field)
(2) algebraic data structures
(Pair, Vector, Matrix, Permutation, Fraction, UPol, Residue)
(3) instances for (2) and related items, in particular, the
algorithmic support for solution of linear systems and such.
I think, the Proposal would suggest for (1) and some part of (0) to
become a part of the proper Prelude.
And (2), (3) have to move to the `support' modules - to be imported
on demand.
Similarly as Num is in Haskell-98 Prelude, and Ratio, `sort'
are in the `support' modules Ratio, List
(if I understand correct the request).
And all the old Prelude remains, except the items like
Num, Integral, Fractional that are replaced by (1).
------------------
Sergey Mechveliani
[EMAIL PROTECTED]
