GNU Go 3.1.12 released October 28, 2001

Teun and Trevor have both recently reported bugs which they
regard as serious and I considered holding up the release
of 3.1.12. However Gunnar looked at Teun's bug and I looked
(briefly) at Trevor's and I don't think there is any strong
evidence that this version is worse broke than (say) 3.1.11.

This includes the following patches:

fuseki9.db.gz. Revised 9 stone fuseki database.
gunnar_1_12.7. Tuning and regression revisions. 
gunnar_1_12.6. Revision of parts of inge_1_12.1. 
trevor_1_12.4. Regression view revisions. 
inge_1_12.1. Combinations. 
gunnar_1_12.5. Readconnect revisions. 
gunnar_1_12.4. Reading code revision. 
trevor_1_12.3. Corrects gg_cputime under unix. 
gunnar_1_12.3. Bugfix for gunnar_1_12.2. 
gunnar_1_12.2. Various changes. 
trevor_1_12.2. Use gg_cputime. 
teun_1_12.3. Does this fix the DFA bug? 
teun_1_12.4. No, but maybe this one does. 
teun_1_12.1. Adds some rotated tests. 
teun_1_12.2. Bugfix in score.c. 
mans_1_12.1. Patch to extract_fuseki. 
gunnar_1_12.1. New and revised regressions. 
trevor_1_12.1. Needed to build on VC.

The patch teun_1_13.1 will have to go in the next release.

- break_chain2_efficient_moves() and break_chain2_moves() revised
- add_array() in readconnect.c simplified
- bugfix in moves_to_connect_in_two_moves()
- sgf traces turned off in quiescence_capture()
- new fields owl_attack_code and owl_defense_code in struct dragon_data
- revised definition of WEAK and WEAKLY_ALIVE dragons in make_dragons()
- revised computation of size of surrounding moyo for a dragon
- 1D revisions
- add_attack_move() and relatives also take the result code as parameter 
- new move reasons ATTACK_MOVE_GOOD_KO, ATTACK_MOVE_BAD_KO, and so on
- move valuation distinguishes between attacks, defenses, owl attacks
  and owl defenses with or without ko
- connections tested as owl defenses when connecting out a critical dragon
- revised valuation of threats and strategical attacks
- compute_eyes_pessimistic() revised
- owl_determine_life() revised
- new static compute_aa_status() broken out of three atari_atari functions
- tuning
- connection tuning
- owl tuning
- joseki tuning
- new and revised regression tests
- new reorientation test suites
- remove rot* subdirs from EXTRA_DIST
- bugfix in score.c (typo from 1D conversion)
- revisions to extract_fuseki: make partial board patterns
- revised fuseki9.db
- Corrected VC build error converting unsigned 64-bit long to double
- regress.pl split into regress.pl & regress.plx
- added -all, to have regress.pl parse Makefile.in for tstfile list.
- html view has better marking of non gg_genmove positions.
- New move generator: combinations() in new file combinations.c
- double attack threats detected
- better valuation of attacks on worms
- general cleanups