This looks really cool. If we could try to implement this as some kind
of Agile Sprint it could be good fun. Looking at http://icfpcontest.org/ it
appears this contest will be significantly different than last years. Since
it looks like you can submit your code to run on any of several (yet
un-named) platforms, I would assume that performance of the application will
not be a significant factor in scoring unlike last year's which was all
about timing. Game simulations seem to be quite popular but there are also
some language or data processing tasks from the prior contests that can be
viewed at links present on the same page.
Hopefully the contest task will be an interesting one.
-- Ben
On Sun, Jun 21, 2009 at 1:17 PM, Kirit Sælensminde <[email protected]>wrote:
> proteus guy wrote:
>
> I'm game. Can you describe what the competition was last time and what you
> did? What languages/tools do you anticipate we'd be using.
>
>
> The task description is here:
> http://smlnj.org/icfp08-contest/task.html
>
> My code is here:
> http://svn.felspar.com/public/icfp/2008/trunk/
>
> Taking last year's as a template it could have been handled by a few
> loosely coupled teams:
>
> - Build and submission -- the entry had to be capable of building and
> running on a VM run by the competition organisers. This meant building a
> build system as part of the submissions and making sure that the builds
> would work on the target environment.
> - Telemetry and rover control -- Not a huge task, but absolutely
> critical.
> - AI -- Where most of the work and the ideas was needed. The rover
> controller would need to provide a good API for this. My AI was the
> simplest
> possible as I didn't have time to put any real work into it.
> - Map generation -- The organisers supplied some simple test maps, but
> to really exercise the AI and control systems more maps would have been
> useful. I didn't make any.
> - A/B AI testing -- Something I didn't even have time to think about
> last year, but some sort of system or procedure to test the AIs against
> each
> other and the maps to make a decision about which should be submitted.
>
> This could all be handled by a small team, or broken out to a few larger
> teams concentrating on each area.
>
> For infrastructure, Felspar can provide a Subversion server and an issue
> tracker/wiki (on our Support site).
>
> If it was just me, I'd probably mainly use C++ and Python -- Python for
> speed of development, and C++ for execution speed. I'd use Boost.Python to
> plug them together. For a larger team though I think it'd be important that
> people could use their preferred languages.
>
> It's almost certainly an early optimisation to try to work out what the
> correct team structure would be and the right tools until we've seen the
> problem. Last year the submission was source code that the organisers had to
> build and run, in most previous years it has been the answer to what amounts
> to a number of complex riddles. Which of these is used this year will make a
> huge difference to how we structure ourselves and what the important tasks
> are.
>
>
> K
>
>
> >
>
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Barcamp Thailand" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/barcamp-thailand?hl=en
-~----------~----~----~----~------~----~------~--~---
