gnugo-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gnugo-devel] easy task to start with


From: Paul Pogonyshev
Subject: Re: [gnugo-devel] easy task to start with
Date: Mon, 18 Oct 2004 23:17:42 +0300
User-agent: KMail/1.5.4

Axel wrote:

> I'd like to participate in the development of GnuGo but I don't know which
> task listed in TODO I should tackle. I'm a good C programmer but a lousy go
> player.

First of all, "GNU Go", not "GnuGo", please.

If you would like to improve the engine itself, I'd suggest starting with
downloading Pike (http://pike.ida.liu.se/) if you don't have it yet.  Then
get yourself familiar with the `regression/view.pike' script, which is
currently the best (and by far) method of tuning GNU Go.  Then pick up a
failing test and try to understand what's wrong with it.  This might be
quite difficult to very difficult.

The best point to start learning the engine is the `engine/reading.c' file
which deals with tactical reading.  It is one of the best modules in GNU
Go and is more or less clean.  Don't forget that GNU Go comes with
documentation.  While some parts of it are not quite up-to-date, it is
still useful to get started with internals.

If you are up to a general maintenance work, try looking at
`patterns/mkpat.c'.  This piece of code is a horrible mess and would
benefit from a full-scale redesign.  This can potentially boggle you down
for weeks, though, as the utility is quite complex.

As for the `TODO', it seems to me that all items are for an experienced
GNU Go hacker.  When I started working with GNU Go, I certainly lacked
skills to implement anything of mentioned there.

Good luck.

Paul





reply via email to

[Prev in Thread] Current Thread [Next in Thread]