[gecode-users] Diagnosing failures

[Malcolm Ryan]

On 13/11/2007, at 7:25 AM, Christian Schulte wrote:

> That's not quite an option: well known, there are too many propagator
> invocations (really, absolutely incomprehensible).

Really? I wouldn't have thought that there were so many. But then  
I've never written an engine.

> Fact is, there has been only a single idea - to my knowledge - how  
> to trace
> propagation at a sufficient level of abstraction, based on so- 
> called S-boxes
> (by Benhamou and others). However it did not make it... We intend  
> to do
> something like that but much later, as a byproduct of some ideas we  
> would
> like to explore (as we are researchers mostly and not software  
> developers,
> so anything we do must be sufficiently new and academically  
> interesting.
> Well...).

I would have thought the issue was really of critical importance to  
making CSPs a really useful programming paradigm. Of course, as an  
academic researcher myself, I realise that there are things that are  
on your research program and things that are not.

> But you have been using Gist, haven't you? That helps quite a lot to
> understand where search goes wrong in my opinion.

I haven't got into that yet. Is there any documentation or examples?

> PS: Was it intentional to have this in private? If not, could you  
> forward
> the message if you agree to that? Thanks.

No, but I think the list has the reply-to-poster flag checked, and I  
don't always remember to CC the list.

Malcolm

> -----Original Message-----
> From: Malcolm Ryan [mailto:malcoonlmr at cse.unsw.edu.au]
> Sent: Monday, November 12, 2007 9:11 PM
> To: cschulte at kth.se
> Subject: Re: [gecode-users] Diagnosing failures
>
>
> On 13/11/2007, at 1:34 AM, Christian Schulte wrote:
>
>> Nothing really, a part from the obvious (and painful):
>
> That's a shame. It would be nice to at least be able to trace the
> order of propagation of constraints.
>
>> decompose your
>> problem into smaller subproblems and check that the subproblems  
>> have a
>> solution.
>
> Fortunately, I have been designing in that fashion from the outset,
> so debugging hasn't been too painful so far. Still, better tools
> would make it easier. Do you have anything planned?
>
> Malcolm
>

--
Many clever men like you have trusted to civilisation.
Many clever Babylonians, many clever Egyptians,
Many clever men at the end of Rome.
                               - G.K.Chesterton, The Napoleon of  
Notting Hill