[gecode-users] Gecode versus Ilog Solver

[Guido Tack]

Hi!

> I want to give some short sentences about my experiences concerning Ilog
> Solver 6.1 and Gecode 1.1 constraint library in C++.

Ok, feedback is always very welcome!

> I formulated my CSP, a special search strategy and some contraint
> propagators both with Ilog and Gecode.
>
> In my opinion the strict parting between modelling level and
> implementation level in Ilog makes its usage complicated. For branching
> or propagation you ve got to switch to implementation level and can use
> some macros to generate the modelling level. but thats no clear solution
>   for me. Your cant get a good interface (e.g. readable headers) for
> your modelling classes.
> The solution of Gecode is better to handle and open for a deeper look
> into details.

That could be stated as one of our design principles ;-)

> As you can see in my often questioning: support is great for Gecode. The
> Ilog support people need at least a week for answering from my own
> experience. That's breaking hard in your programming progress. Thanks
> for your support! ;) (looking forward to your tutorial.. that will deal
> with the remaining starting problems!)

You're very welcome. Of course, one difference is that we can't give any 
guarantees that we're going to respond at all, while Ilog pays people for 
giving support. Of course, we'll try to do our best.

Concerning the tutorial, in fact, people asking questions gives us a good idea 
of what parts need special care. So, any feedback on what you consider 
especially difficult (or fairly easy) is welcome.

> Summarising I can say that my developing and modelling time in Gecode is
> at least half than in Ilog. Maybe in part by a better knowledge of some
> techniques in such libraries now (tried Ilog first).
>
> But in the end Gecode is so far easier to handle for me.. ;)
>
> Think some enjoyable words are also nice for such a user list!

Indeed they are! Thanks a lot. And keep bothering us ;-)

Cheers,
	Guido

-- 
Guido Tack
Programming Systems Lab, Saarland University, Germany
http://www.ps.uni-sb.de/~tack