Re: [eclipse-users] Java exception is being thrown when invoking an eclipse program scheduling over 300 tasks

Hi Kish,

This was extremely helpful to me.  I changed my code and did increase
the efficiency of my algorithm.  My question now however is the
get_domain function.  I try this:

ic_sets:(T1::[]..[1,2,3,4,5]), writeln(T1), T2 in T1, writeln(T2),
ic:(get_domain(T2, T3)), writeln(T3), ic:(T4::1..5), writeln(T4).

The ideas is to turn T1 into an ic dom variable.  The output of this program is:

T1{([] .. [1, 2, 3, 4, 5]) : _604{0 .. 5}}
T2{1 .. 5}
1 .. 5
T4{0 .. 5}

It appears that T3 and T4 are not the same kind of variable.  Is that true?

-Tim

P.S.

My code is evolving at:

http://adempiere.svn.sourceforge.net/viewvc/adempiere/branches/libero/extend/src/test/functional/fcsched_test.pl?view=markup

Please feel free to make other suggestions!  I can't even tell you in
words how grateful I am for this help.

On Dec 12, 2007 6:06 PM, Kish Shen <kisshen_at_...5...> wrote:
> Hi,
>
> William Heath wrote:
> > Hi Kish,
> >
> > I figured it out and it is definitely a memory issue with eclipse-clp.
> >  What I did to solve it was this:
> >
> >   java.lang.System.setProperty("eclipse.global-size", "1024");
> >   java.lang.System.setProperty("eclipse.local-size", "1024");
> >
> I should have added that I think it is unlikely that you require
> expanding the local stack size -- all the domains and lists are stored
> in the global stack (as specified by global-size).
> >
> > convertIC_SET_to_IC_DOM(Ic_set, Ic_dom):-
> >           writeln('convertIC_SET_to_IC_DOM'),
> >               ic_sets:(Ic_domVar in Ic_set),
> >               ic:(get_domain_as_list(Ic_domVar,Ic_setIntegerList)),
> >               ic:(Ic_dom::Ic_setIntegerList).
> >
> >
> You seem to be trying to copy the ic domain from one variable to
> another. I am not quite sure why you are doing this: why can't you use
> Ic_domVar directly? In any case, using get_domain_as_list/2 is very
> likely to be much less efficient than using get_domain/2, unless you
> have a large number of holes (and given your domain size, a *very* large
> number of holes, somewhere in the same order of the size of your domain).
>
> But I think this is somewhat secondary, as I  said in my last message,
> using such a large domain is probably quite inefficient.
>
> Cheers,
>
> Kish
>