FD.exactly made avare of '\\=:'

[Christian Schulte]

Hi Gabor,

your idea sounds pretty reasonable to me, although quite nontrivial. But
you are true, you have to resort to doing symbolic reasoning as the
constraint store is not capable of expressing disequalities.

As to the priorities (I found it after some code digging): the
propagators are served by a thread that runs with normal priority. So if
you create a thread with higher priority, then you will a greater share
of computation time that the propagators will get. However, it does not
mean that you can ensure that your code is run before any of the
propagators is run.

Christian

"Gabor Szokoli" <gabor.szokoli at vanderbilt.edu> schrieb im Newsbeitrag
news:<091d01c19eb2$ed4b31e0$fd693b81 at szocske>...
> Hello!
> 
> I'm back... ;-)
> 
> I have a rather poorly propagating FD search problem:
> I have lots of FD.exactly propagators on lists with lots of
> inequalities posted among list members. (mostly within the list)
> I'd like the upper bound on the counter variable to take the
> inequalities into account for example.
> (largest fully connected graph in the complementer of the inequality
> graph comes to mind first as an approximation)
> Has anyone ever experimented with anything simlar?
> I'm thinking along the lines of passing the \=: constraints
> represented as listindex-pairs to this new propagator. (I don't feel
> like rewriting the whole FD library to add a new domain dimension and
> wake-up events before giving it a try first :-) )
> 
> I'm still interested in work on propagator precedence. Is it true that
> thread priorities are effective to have precedence among
> propagator-mimicing threads with FD.watch.X in them?
> If yes, what priority do real propagators have compared to these?
> (higher than 'high' I suppose)
> 
> 
> Thanks:
> 
> Gabor Szokoli
> www.isis.vanderbilt.edu
> 
> -
> Please send submissions to users at mozart-oz.org
> and administriva mail to users-request at mozart-oz.org.
> The Mozart Oz web site is at http://www.mozart-oz.org/.


-
Please send submissions to users at mozart-oz.org
and administriva mail to users-request at mozart-oz.org.
The Mozart Oz web site is at http://www.mozart-oz.org/.