[gecode-users] order of "wait" statements maintained?

[Christian Schulte]

Oh, don't worry about performance... That's miniscule compared to
propagation.

Branchers are always executed, even if all variables are assigned (but the
might not generate any branches then).

Cheers
Christian

--
Christian Schulte, Professor of Computer Science, KTH,
www.ict.kth.se/~cschulte/



-----Original Message-----
From: Martin Mann [mailto:mmann at informatik.uni-freiburg.de] 
Sent: Tuesday, April 30, 2013 10:27 AM
To: cschulte at kth.se
Cc: 'gecode user list'
Subject: Re: [gecode-users] order of "wait" statements maintained?


Another question arised when posting this


	branch(*this, A, ..);
	branch( *this, &checkA );
	branch(*this, B, ..);
	wait( *this, A+B, &checkB );

I thought about replacing the wait with another branch call. But are
branchers always called even if all variables are assigned?
Or is the given solution better than the "4-branch-option"?

And where is the true difference also for performance etc.? Guess the wait
call does more checks on the domains?!?

Thanks,
Martin


Am 30.04.2013 09:58, schrieb Martin Mann:
>
> Hi Christian,
>
> guessed as much.. mhh.. think the additional bool var wont be a 
> problem but I will give the branch call with &checkA a first try, 
> since it keeps the CSP small without further variables etc.
>
> Thanks for you help,
> Martin
>
>
> Am 29.04.2013 20:06, schrieb Christian Schulte:
>> Hi Martin,
>>
>> There is no order among several waits. You could either use branch 
>> with a function as you suggested (calling status is in fact fine) or 
>> you could encode the order yourself. Make checkA and checkB assign a 
>> variable and then you wait on these variables as well. Would that do?
>>
>> Cheers
>> Christian
>>
>> --
>> Christian Schulte, www.ict.kth.se/~cschulte/
>>
>>
>>> -----Original Message-----
>>> From: users-bounces at gecode.org [mailto:users-bounces at gecode.org] On 
>>> Behalf Of Martin Mann
>>> Sent: Monday, April 29, 2013 5:39 PM
>>> To: gecode user list
>>> Subject: [gecode-users] order of "wait" statements maintained?
>>>
>>>
>>> Hi everybody,
>>>
>>> for some processing, I post something like this in my Space 
>>> constructor
>>>
>>>     wait( *this,   A, &checkA);
>>>     wait( *this,   B, &checkB);
>>>     wait( *this, A+B, &printAB);
>>>
>>>     branch( *this, A, .. );
>>>     branch( *this, B, .. );
>>>
>>> to do some dedicated successive checks on assignments of A, B and 
>>> both A
>> and
>>> B.
>>>
>>> Since my last printAB depends on the outcome of both checkA and 
>>> checkB
>> (both
>>> might make the CSP fail) I need to know if the processing order of 
>>> the
>> wait
>>> statements the same as their posting order within the constructor... 
>>> :/
>>>
>>> Or is there a better solution to that as posting an according 
>>> branching
>> like
>>>
>>>     branch( *this, &checkA );
>>>
>>> where I first have to trigger constraint propagation in order to 
>>> ensure assignment... !??!
>>>
>>> Thanks for your help,
>>> Martin
>>>
>>>

--
Dr. Martin Mann, Postdoc
Bioinformatics - Inst. of Computer Science Albert-Ludwigs-University
Freiburg
Tel: ++49-761-203-8254
Fax: ++49-761-203-7462
http://www.bioinf.uni-freiburg.de/~mmann/