[ANN] CloCoP - constraint programming for Clojure

[Alex Engelberg]

http://github.com/aengelberg/clocop

CloCoP is a Clojure wrapper of the Java library JaCoP. The acronyms stand for "Clojure/Java Constraint Programming". This invites comparison to the core.logic library, and you may wonder why we need both. There are a few ways in which, in my opinion, the JaCoP system is better than core.logic:

• JaCoP is more "plug-in-able," with an extensive set of customizations to the way that the search operates. There are interfaces for different components of the search, and each has several implementations.
• I found that with core.logic, I was somewhat limited by the set of available constraints. JaCoP has many different constraints that seem to more suit my needs for solving challenging problems.
• As the core.logic people say, JaCoP is anywhere from 10X-100X faster than core.logic at solving Finite Domain problems.

JaCoP has a lot of "global constraints" which are very powerful and essential for describing certain problems. As Radoslaw Szymanek (an author of JaCoP) says, "CP without global constraints is just [a] plain academic toy. Using problems with arithmetic constraints is doing CP bad publicity."

If you'd like to see implementations of sample problems in CloCoP, check out the test cases (https://github.com/aengelberg/clocop/tree/master/test/clocop).

[David Nolen]

Nice work! :)

--
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clo...@googlegroups.com
Note that posts from new members are moderated - please be patient with your first post.
To unsubscribe from this group, send email to
clojure+u...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
---
You received this message because you are subscribed to the Google Groups "Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email to clojure+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[sesm]

Great stuff!
Unfortunately, README doesn't say anything about license and contributing, so I've sent a pull request to check it :) If you accept contributing tests/examples, I would send a lot more.

вторник, 10 сентября 2013 г., 5:39:46 UTC+4 пользователь Alex Engelberg написал:

[Alex Engelberg]

It's under the Eclipse Public License (as stated in project.clj), with pull requests welcome. I just added the license info to the Readme as well. Sorry about the confusion.

I've also created a 0.2.1 branch (without any changes yet) for people to pull-request new features into. (Though if you're just submitting examples/tests, the master branch is fine.)

Thanks,

--Alex

[Daniel Jomphe]

So then you'd also need a 0.3.0 branch to allow people to pull-request new features and/or breaking changes, since those things are not semantically ok to do in a patch release (the 1 in 0.2.1).

Of course there are many other ways to look at this workflow thing around pull requests, versioning and etc. I hope my tone felt helpful. :)

[Dmitry Groshev]

Great stuff!

I'm wondering what's the "realworld" difference between JaCoP and CHOCO. Why did you choose the former?

[Alex Engelberg]

To Daniel's comment: that's a great idea (I'm sorta new to this whole releasing-a-library thing so I didn't think of that), and I'll definitely make a branch for that if I think of a big idea that I want to implement that involves breaking changes. But I figure that if someone other than me thinks of something like that, we'll probably discuss it a bit and figure something out (rather than just showing up with a pull request).

To Dmitry's comment: I've actually never tried Choco, I essentially chose JaCoP because it occured first in my "Java Constraint Programming" google search. Now I'm seeing that a lot of constraint programmers use Choco, and I'll try it out. I will point out that JaCoP walked away with some bronze and silver medals (with Gecode getting ALL the golds) at the FlatZinc challenge. (Choco didn't compete at all, though.)