[gnugo-devel] GNU Go 3.1.23

[Daniel Bump]

GNU Go 3.1.23 is now available at:

http://www.gnu.org/software/gnugo/devel.html

As I explained earlier, there are two outstanding patches that 
I wanted to get in but when I saw the favorable regression results
it seemed like a good time to do the release. See regression/BREAKAGE.

heikki_1_23.3. New GTP functions.
arend_1_23.4. Experimental influence. 
arend_1_23.5. Owl threats. 
inge_1_23.4. Removal of DEFEND_BOTH_MOVE. 
gunnar_1_23.5. Test revisions. 
gunnar_1_23.4. Two groups in hash key. 
trevor_1_23.3. Test corrections. 
arend_1_23.3. Hoshi tuning. 
teun_1_23.3. Don't check constraint diagram when patno == 0 
trevor_1_23.2a. Regression view enhancements. 
gunnar_1_23.3. Bugfixes in worm.c and value_moves.c. 
readconnect_1_23.1. Connection through ponnuki and more. 
inge_1_23.3. Revaluation of some worms. 
trevor_1_23.1.tar.gz. Makes VC run the DFA. 
teun_1_23.2. Turns up some array bound violations. 
teun_1_23.1. Pattern consistency checking. 
gunnar_1_23.2. Tuning and pattern bugfixes. 
gunnar_1_23.1. Atari_atari revision. 
inge_1_23.2. ALL_MOVE. 
inge_1_23.1. Removal of ATTACK_EITHER_MOVE. 
arend_1_23.2. NNGS tests. 
brandsma_1_22.1. VC revision. 
teun_1_22.1. Revised config.vcin. 

- configure option: alternate-connections
- experimental influence: followup value computed in influence.c
- experimental influence: new_value_territory revised
- experimental influence: new handling for intrusion patterns
- experimental influence: support for standard/experimental specific patterns
- reevaluate_ko_threats now makes checks
- new get_biggest_owl_target()
- DEFEND_BOTH_MOVE reason taken down
- new gtp functions get_komi and get_handicap
- bugfix in gtp worm_stones: did not print newline if board empty
- support for two strings and two results in cache.[ch]
- simplified code to do cache lookup in reading.c and owl.c
- bugfix in estimate_territorial_value()
- bugfix in find_worm_threats()
- readconnect finds connections through ponnuki
- readconnect sgf improvements
- bugfix in estimate_strategical_value(): ALL_MOVE worms were overvalued
- VC build now DFA compliant
- semeai module sets matcher status to ALIVE instead of ALIVE_IN_SEKI
- regression html view enhancements
- fix arraybound violation in compute_aa_values
- constraint diagram checking moved to new check_constraint_diagram()
- Invoke check_constraint_diagram *before* incrementing patno
- atari_atari_find_defense_moves() revised
- ATTACK_EITHER_MOVE taken down
- new ALL_MOVE reason
- new static find_all_data(), all_move_redundant() in move_reasons.c
- new function add_all_move()
- add_defend_both autohelper macro revised
- tuning
- influence tuning
- owl tuning
- hoshi tuning
- new and revised tests 

Dan