[boost] Spamjunk Set Library --> Finite Domain Constraints

From: David Bergman (davidb_at_[hidden])
Date: 2002-08-24 15:28:49

I have a proposal, which necessitates a shift of conception. 

First, a note regarding "powerset":

The powerset term does make *some* sense, even for the "math" guys out
there, since we here have a (very) finite set as the universe. So, even
though the individual sets are not powersets, the operations take place
in a very well-defined powerset. This in contrast to the set concept
realized by the STL set template, in which the universal set is not
given (beside the implicit limits of the type T representation,
obviously) beforehand.
But, since the actual sets are not powersets (or even general sets of
sets), there might be better alternative nomenclature to be found in the
domain of Constraint Programming (and, more specifically, in Constraints
Over Finite Domains), which is how I tend to conceive the "Spamjunk" Set
Library.

The proposal:

We change our conceptualization a bit, and use the "Spamjunk" Set
Library as the foundation of our own Boost.FiniteDomainConstraint
library (no, Boost.FDC would be a bit too cryptic ;-).

The easiest form of constraint is a simple enumeration of the elements
satisyfing that constraint, which is realized already by the "SpamJunk"
Set Library. We can hold on with defining further restricting constructs
till later. So we would have a poor man's Finite Domain Constraint
system initially...

Till we have a sufficiently versatile FDC solution, we could always
borrow the term "Domain", and have the "Spamjunk" Set Library be
Boost.Domain, with the clear "constraint" interpretation of the term
"domain".

/David