[gnugo-devel] GNU Go 3.1.26

[Daniel Bump]

This release makes the experimental influence, experimental
connection and alternate connection configure options standard.
To get the old behavior configure --disable-experimental-influence etc.

A lot of test results are changed without clear improvement.
Yet other evidence is that the experimental influence option
makes GNU Go a lot more aggressive and stronger.

I have conflicting results whether this release is faster
than 3.1.25 or not. It's possible that it's a lot faster.
That would be due to the patch arend_1_26.3.

gunnar_1_26.6. Caching code repair. 
trevor_1_26.1. makevcdist.pl changes. 
arend_1_26.5. Owl Patterns with value 99 are handled correctly. 
gunnar_1_26.5. Code cleanup. 
arend_1_26.3. Reorganization of owl callback, promising a big speedup. 
tuning_1_26.1. A couple of edge joseki patterns. 
gunnar_1_26.4. Pattern cleanup. 
gunnar_1_26.3. Spelling corrections. 
arend_1_26.4. Verbosity. 
scoring_1_26.1. Bugfix in close_bubbles. 
autolevel_1_26.1. Bugfix in autolevel 
gunnar_1_26.2. Semeai bugfix. 
gunnar_1_26.1. Revised "stepped on a bug" message. 
arend_1_26.2. Use old influence for deciding which owls to examine.
trevor_1_26.1. Alternative non-variadic trace. 
arend_1_26.1. Experimental influence speedup. 
trevor_1_25.4. Stack assertion. Arrived too late to go in 3.1.25. 

- experimental influence, experimental and alternate connections are default
- caching code revised to once more work with TRACE_READ_RESULT enabled
- added EXPERIMENTAL_READING to config.vc(in)
- updated makevcdist.pl
- bugfix in owl.c:dump_pattern_list() 
- alternative reorganized owl callback structure for substantial speed-up:
- owl_shapes revised, new functions get_next_move_from_list() et al.
- new function dump_pattern_list(...) for debugging
- pattern cleanup
- spelling fixes
- verbosity revisions with alternate connections in readconnect.c
- bugfix in close_bubbles()
- bugfix in compute_owl_escape_values()
- revised "stepped on a bug" message
- bugfix in update_status in semeai.c
- bugfix in estimate_time_by_move
- replay added to debug helpstring
- revised nonvariadic versions of TRACE and allies
- increased MAXSTACK to MAX_BOARD*MAX_BOARD again
- experimental influence taken down for initial influence when dragons unknown
- doc revision

Dan