[gecode-users] Gecode 3.0.0: What we will do. What do you want?

[Christian Schulte]

Dear David and Malcolm,

thanks for the nice and kind remarks. Highly appreciated.

Let me please go off a little bit into today's philosophy lecture (I know
it's another bore). One thing that we completely resist in Gecode is
anticipating the future (as in what could potentially be useful). We do what
we understand and where we also know it will be used and how it will be
used. That we try to do as simple, general, and efficient as possible. We
never do guesswork, if it does not have an application right now we won't do
anything. Otherwise one will guess and one will always make the wrong guess.
I think this commitment to simplicity is one of the really strong aspects of
Gecode. No fluff, no code infected with bitrot, nothing. I know it sounds
heretic but my now 15 years of system building experience has taught me that
lesson.

Of course, what we do in our research is something else again.

But if you have a clear picture of what it is what you want we will for sure
try to come up with something (here the "we" could of course include you)!
The scenario of hacking the system is definitely not anything we want!

I wrote down Malcolm's cry for better tools. It's a pity, we know something
that should do exactly what Malcolm desires but we do not have the
resources... But we will try harder to find the resources.

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 David Rijsman
Sent: Thursday, June 26, 2008 11:30 AM
To: users at gecode.org
Subject: Re: [gecode-users] Gecode 3.0.0: What we will do. What do you want?

I fully agree with Malcolm, on all remarks! 

I really like the open structure of Gecode, it is extremely easy to
interface to another system but it seems that for debugging and
visualization this is no longer valid. It is great to have gist but I
require to have my own information and visualization both for the logic as
the control. For logic Gecode is open enough for me to create information on
the constraint network and also on domain reductions after adding
constraints, but for the control I had no other option than to add code into
the existing code base. This is not really what you want in an open system,
I like to add code but not into existing files.  I will potentially have to
redo my work for every release (especially 3.0.0 will have new branching and
lots of more control). 

- What would be nice is to have is access to events in the control, with
appropriate data, such that I can use this in my own system.

hope this makes sense,


David J Rijsman
Software Architect
Quintiq
 
T +31 (0)73 691 07 39
F +31 (0)73 691 07 54
M +31 (0)62 127 68 29
E david.rijsman at quintiq.nl
I www.quintiq.com


>>> 
From: 	Malcolm Ryan <malcolmr at cse.unsw.edu.au>
To:	gecode list <users at gecode.org>
Date: 	6/26/2008 10:26 AM
Subject: 	Re: [gecode-users] Gecode 3.0.0: What we will do. What do
you want?

On 26/06/2008, at 2:07 AM, Christian Schulte wrote:
> Now: What do you want? How can you help us? Are we missing something  
> really
> important? Is there something fundamentally wrong?

Firstly, let me thank you for a fabulous tool and better support than  
any free software I've encountered before. Research software is not  
usually any near the quality of Gecode and you deserve high acclaim  
for your work.

As for improvements, in the short term I think the push to fill out  
the documentation is important. In the longer term, it would be good  
to see more visualisation and debugging tools. If you make an error in  
your constraint definitions, at the moment it is very difficult to  
detect and isolate the problem. It is also hard to visualise whether  
your constraint system is propagating and searching efficiently. It  
would be very nice to have a graphical way to watch the propagation of  
information around the constraint graph.

Like any programming language, there are good and bad ways of writing  
constraint systems. At the moment the modelling problem remains a  
black art, only really properly understood by a few. I think it will  
remain that way as long as constraint propagation remains a black box.  
Solving this may require a significant about of HCI work, but I think  
it would be very valuable in the long run.

Malcolm

_______________________________________________
Gecode users mailing list
users at gecode.org 
https://www.gecode.org/mailman/listinfo/gecode-users


This message contains information that may be privileged or confidential and
is the property of Quintiq. It is only intended for the person to whom it is
addressed. If you are not the intended recipient, you are not authorized to
read, print, retain, copy, disseminate, distribute or use this message or
any part thereof. If you have received this message in error, please notify
the sender immediately and delete all copies of this message. Please note
that e-mails are susceptible to change, therefore they are not binding.

_______________________________________________
Gecode users mailing list
users at gecode.org
https://www.gecode.org/mailman/listinfo/gecode-users