| Make way for new thrust of HouseBot development |
|
More...
|
over 16 years ago
|
| A minimal komi gtp command implementation. Just enough to be of use in the loadSgf method but not really allowing to set the komi to anything but 7.5 (as before). |
|
More...
|
over 16 years ago
|
| SGF unit test no longer leaves lots of threads lying around. (Functions operating on 'mT' were being done in the wrong thread, and quit therefore wasn't working properly) |
|
More...
|
over 16 years ago
|
| Also ignore *.log files |
|
More...
|
over 16 years ago
|
| Implemented B, W, AB, and AW and wrote initial tests |
|
More...
|
over 16 years ago
|
| Avoid situation where search agent is not alive (AKA, was told to shut down), yet is waiting for more commands |
|
More...
|
over 16 years ago
|
| * Implemented and tested playing a black move from an SGF file |
|
More...
|
over 16 years ago
|
| Use playerAction instead of goAction |
|
More...
|
over 16 years ago
|
| SGF to GTP move conversion |
|
More...
|
over 16 years ago
|
| loadSgf is now the method used for GTP and _loadSgf the internal one (also for testing). The delegate didn't work (it segfaulted). |
|
More...
|
over 16 years ago
|
| First test case for loadsgf (and passing implementation thereof) |
|
More...
|
over 16 years ago
|
| Fixed eye point detection |
|
More...
|
over 16 years ago
|
| Made goGameDefinition.toString() more readable |
|
More...
|
over 16 years ago
|
| * Implemented the showboard GTP command |
|
More...
|
over 16 years ago
|
| Basic diagonal tracking (A precursor to 3x3 patterns). 8% slowdown |
|
More...
|
over 16 years ago
|
| Replaced simple properties with raw member variables. 10% speedup |
|
More...
|
over 16 years ago
|
| Short circuit solution status calculations for the common cases |
|
More...
|
over 16 years ago
|
| Avoid useless struct copies of goGameStateAtPosition. 4% speed up |
|
More...
|
over 16 years ago
|
| What do you know, my computer hates long division too! 5% speedup |
|
More...
|
over 16 years ago
|
| Removed use of dynamic array in RAVE simulations. 15% speedup |
|
More...
|
over 16 years ago
|
| Fixed overflow bug with integer arithmetic and got % of runtime down for confidenceBound and mean functions. |
|
More...
|
over 16 years ago
|
| 2% speedup through faster confidence bound calculations |
|
More...
|
over 16 years ago
|
| Finally... the compiler workaround to let HouseBot compile with dmd again |
|
More...
|
over 16 years ago
|
| Removed trait code from tangoBind.d and used Tango's instead |
|
More...
|
over 16 years ago
|
| One step closer to having dmd support again (dmd currently segfaults but no longer issues warnings) |
|
More...
|
over 16 years ago
|
| Foundation for random weighted moves (more to come). |
|
More...
|
almost 17 years ago
|
| Converted uct to use monte carlo player |
|
More...
|
almost 17 years ago
|
| Moved the common logging to tango.util.log. Default log level is now higher and logging goes to housebot.log. This includes new output for time control which is intended for future plotting of time tracking. |
|
More...
|
almost 17 years ago
|
| Set default search mode to UCT+RAVE since it is much stronger |
|
More...
|
almost 17 years ago
|
| First idea for an implementation of the loadSgf command |
|
More...
|
almost 17 years ago
|
