gnugo-devel
[Top][All Lists]
Advanced

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

[gnugo-devel] mostly 1D-patch


From: Gunnar Farneback
Subject: [gnugo-devel] mostly 1D-patch
Date: Tue, 16 Oct 2001 21:36:21 +0200
User-agent: EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode)

I've put up a big patch at
http://www.lysator.liu.se/~gunnar/gnugo/patches/gunnar_1_11.4.gz

- 1D changes
- cleaning up
- new debug flag DEBUG_READING_CACHE
- same_dragon() renamed is_same_dragon()
- new gtp commands start_sgftrace and finish_sgftrace
- new and revised regression tests

This is mostly a lot of 1D stuff and a little general cleaning. The
most interesting part are the new gtp commands start_sgftrace and
finish_sgftrace. With these you can do fancy things like

loadsgf golois/Handtalk980821-1.sgf
trymove black D4
trymove white C4
trymove black E4
trymove white C5
trymove black C3
trymove white C2
trymove black B3
start_sgftrace
attack B3
finish_sgftrace vars.sgf
popgo
popgo
popgo
popgo
popgo
popgo
popgo

which lets you generate an sgf dump of the tactical reading of a
string deep down in some other reading problem. However, if you make a
mistake there's a definite risk that you either get a useless result
or flood your RAM with a huge in-memory sgf tree.

Hmm, this reminds me that finish_sgftrace had better call the
relatively new functions to free the memory used by an sgf tree. This
will come in a later patch.

Somewhere in the 1D changes I seem to have fixed a reorientation
problem since two of the owl_rot tests unexpectedly pass. There are
also a few other unexpected result (both failures and passes) which
are probably related. I've checked the failures and it seems likely
they are all caused by owl reading results that have changed.

This patch is not yet in CVS but I will add it shortly.

/Gunnar



reply via email to

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