bsamseth / Goldfish / 287

Builds Branch Commit Type Ran Committer Via Coverage
287 search-refactor Add optional debug logging to puzzle tests. Disabled in all public commits, but there for easy enable/disable when debugging. push bsamseth travis-ci pending completion   set done
272 search-refactor Update puzzle tests to play nice with unittest Pull #27 web-flow travis-ci pending completion   set done
271 search-refactor Update puzzle tests to play nice with unittest push bsamseth travis-ci pending completion   set done
269 search-refactor Clean up bound determining in search push bsamseth travis-ci pending completion   set done
270 search-refactor Clean up bound determining in search Pull #27 web-flow travis-ci pending completion   set done
268 search-refactor Bugfix: Be more restrictive about using NMP Null-move pruning was previously restricted to be used when beta <= CHECKMATE, which essentially was no limit at all. We only do NMP for a potential beta-cutoff, i.e. the null-move-score is >= beta. But... Pull #27 web-flow travis-ci pending completion   set done
267 search-refactor Bugfix: Be more restrictive about using NMP Null-move pruning was previously restricted to be used when beta <= CHECKMATE, which essentially was no limit at all. We only do NMP for a potential beta-cutoff, i.e. the null-move-score is >= beta. But... push bsamseth travis-ci pending completion   set done
266 master Update default match script settings. Changed to allow _way_ more end game play. This will allow engines with improvements in this area to hopefully shine through in the testing. It will cause the matches to take _way_ longer of course, but that ... push bsamseth travis-ci pending completion   set done
265 master Add missing include for std::abs Suddenly failed to compile with gcc 6. Explicit include fixes it. push bsamseth travis-ci pending completion   set done
264 master Fix minor LGTM issues. push bsamseth travis-ci pending completion   set done
263 master Bugfix: Don't save quiescent results in ttable (#25) I had originally thought that storing quiescent results with a depth of zero would be safe (motivated by imporved benchmark scores (not noticing the changes in evaluations...)). Turns out, this... push bsamseth travis-ci pending completion   set done
262 v1.8.2 Bugfix: Don't save quiescent results in ttable (#25) I had originally thought that storing quiescent results with a depth of zero would be safe (motivated by imporved benchmark scores (not noticing the changes in evaluations...)). Turns out, this... push bsamseth travis-ci pending completion   set done
261 master Stop searching deeper when optimal mate is found. If we have searched to a depth d, and evaluated the position as a mate in <= d plies (for either side), then nothing at a deeper level will change that evaluation. So we can safely stop the search... push bsamseth travis-ci pending completion   set done
260 master Debug with safe optimizations push bsamseth travis-ci pending completion   set done
259 v1.8.1 Mark version v1.8.1 as merge of aspiration and minor tweaks. push bsamseth travis-ci pending completion   set done