Re: [gnugo-devel] Readconnect

[Daniel Bump]

Tristan wrote:

> > Maybe we should link the code in right now on an experimental
> > basis, the same way the new semeai code is experimentally
> > linked in now.
> 
> I do not know, I do not even know where it should be linked,
> and what is exactly expected from it.
> Linking it to the engine and testing it even experimentally
> is needed now I think (the code can now read non trivial
> connections, and it is interesting to see what the engine
> does with it).

We need to replace find_connections() and find_cuts() which
are called from make_dragons.c. Everything should be
contingent on a variable experimental_connections which is
handled exactly like experimental_semeai, so that the
experimental connection code can be ignored by default until
we can tell if it's an improvement. You would get an
experimental version by running 'configure
--enable-experimental-connections' then recompiling.

It seems to me that we should continue to use the pattern
matcher to find candidates dragons then use string_connect()
to decide whether the match is valid. If Gunnar agrees with
this plan, probably it is easiest if we implement it. Gunnar
might have a different idea of how it should work, though.

Gunnar, is this how you'd do it?

> Is it possible to test readconnect on these mistakes ?
> Where can I find them ?

Connection mistakes happen frequently. I would say
in most games. A typical example is trevor test 9:

http://www.public32.com/regress/?trevor:9

GNU Go thinks the white stones are all one dragon,
which you can see from trevor's diagram since they
are all labelled by the same letter z. You can
tell at a glance from trevor's web page that this
is a connection problem, which is one reason Trevor's
html views of the regressions are so useful.

Dan