
Boost : 
From: Joel Young (jdy_at_[hidden])
Date: 20020904 14:08:15
From: "David Bergman" <davidb_at_[hidden]>
> I underline Joel's sumbission.
Thanks David,
From: "David Bergman" <davidb_at_[hidden]>
> I totally agree. Why complicate matters with obscure orderings and
> tribools, when we have quite concrete elements to work with.
This is a pieinthesky vision I have.
Take a pair of types <t1, t2> and take a set of binary boolean relations
on (t1, t2) and create a new type which manages the application of these
relations to those types.
For example, say I had <interval, interval> and the 13 relations, I could say:
define I as a set of intervals.
define R as a set of relations.
now return the subset G of intervals of I such that
for all g in G, g in I .and. g R g
and other tasks such as transitive closures, fixpoints, etc.
One can express constraint systems:
F, G, H, I be sets of intervals
P, Q, R, S be sets of relations
return me
set of all tuples (f, g, h, i) over (F, G, H, I) such that
f P g .and. g Q h .and. h R i .and. i S f
Does this make any sense?
Joel
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk