Re: [gnugo-devel] GNU Go 3.3.3

[Trevor Morris]

Dan:
Did you update the regression test results?  I updated 
http://www.public32.com/regress/
and there are a lot of regression differences.
-Trevor


At 02:17 PM 6/4/2002 -0700, you wrote:

>GNU Go 3.3.3 exists. Get it at:
>
>http://www.gnu.org/software/gnugo/devel.html
>
>dan_3_3.1. New test cases. 
>gunnar_3_3.10 Cleanup. 
>gunnar_3_3.9 Experimental connections. 
>trevor_3_3.2. DFA compile time speedup. 
>arend_3_3.7. Experimental influence no longer experimental. 
>gunnar_3_3.8. Tuning. 
>arend_3_3.6. Less owl pushing, big speedup promised. 
>arend_3_3.5. DFA handling of symmetric patterns. 
>escape_3_3.1. Don't try too hard to make eyes while escaping. 
>arend_3_3.4. Reading cleanup. 
>gunnar_3_3.7. Trace and debug improvements. 
>arend_3_3.3. Functions in reading.c renamed. 
>gunnar_3_3.6. Experimental connections no longer experimental. 
>gunnar_3_3.5. CB242 reverted. 
>gunnar_3_3.4. Dragon field in struct eye_data removed. 
>gunnar_3_3.3 Cleanup. 
>gunnar_3_3.2 3.0 FIXME notices removed. 
>gunnar_3_3.1. hoshi.sgf repaired. 
>arend_3_3.1b. Revision of the following patch. 
>arend_3_3.1. Various. 
>dragon_data_3.3.1. New --decide-ragon-data command line option. 
>owl_nodes_3_3.1. Tuning patch. 
>tuning_3_3.1. Tuning patch. 
>
>- new function owl_strong_dragon() and new autohelper with the same name
>- vital_chain() returns the max number of eyes
>- no amalgamation around eye spaces with experimental_connections enabled
>- eye spaces no longer modified by INHIBIT_CONNECTION with
>  experimental_connections enabled
>- bugfix in find_connection_moves()
>- pattern classes Y and W introduced in conn.db to identify patterns
>  which only applies with or without experimental_connections
>- patterns with class e in conn.db changed to class I to better
>  reflect current usage
>- connection tuning
>- improves DFA compile time by doing fewest possible large syncs 
>- experimental_influence command line and configure option removed
>- new flags dragons_known and is_territorial_influence in influence_data
>- struct local_owl_data for speedup only holds eye space of relevant color
>- owl_eyespace() and allies use only one eyespace
>- dfa doesn't report symmetric patterns multiple times
>- escape patterns in owl_defendpats.db given E classification
>- owl_determine_life() not called during escape attempt
>- only print debug output for eyes during owl reading if owl debug is on
>- better traces for unused owl patterns
>- --decide-connection also reports ko results
>- bugfix in load_and_score_sgf_file()
>- new test cases
>- edge_block and double_atari_chain2 renamed to *_moves
>- do_attack uses simple_ladder_attack when stackp > depth+10
>- give up simple_ladder_attack below 20 ply when we are komaster
>- unused caching code removed from attack? and defend?
>- macros used to make reading code readable
>- dragon field in struct eye_data removed
>- new function find_eye_dragons() in optics.c
>- dragon_invincible() and genus computation in make_dragons() revised
>- play_connect_n() uses string_connect() and disconnect()
>  unconditionally, no longer contingent on experimental_connections
>- default value for experimental_connections reset to 0
>- new command line option --decide-dragon-data
>- consistency checking of replacement patterns reenabled for unstable line
>- tuning
>- new tests and sgf files
>
>Dan
>
>_______________________________________________
>gnugo-devel mailing list
>address@hidden
>http://mail.gnu.org/mailman/listinfo/gnugo-devel