diff options
| author | Scott Gasch <[email protected]> | 2026-10-14 12:19:02 -0700 |
|---|---|---|
| committer | Scott Gasch <[email protected]> | 2026-10-14 12:19:02 -0700 |
| commit | 78a5ea00ec507164e1cfc7fd12846e1b05d78e7e (patch) | |
| tree | 4c35bc24c59fdc5f30389db371b732da3b376d70 /src/genetic/._main.cpp | |
| parent | ac5cc078b71e94b21a8a843628045c745022aa75 (diff) | |
Remove PRUNE_LIVE -- never wired to anything, and separately broken
PRUNE_LIVE=<K> was meant to actually prune moves below a predicted-FH
threshold for branching-factor measurement, but PruneLivePredict (the
macro alias for search.c's PruneVerifyPredict) was never called from
anywhere -- confirmed by grep: its only two references in the file
are its own #define and declaration. The wiring to make PRUNE_LIVE=<K>
actually prune a move was never written, so this flag has done
nothing even in builds where it compiled.
It also didn't compile on its own: PruneVerifyPredict returns
g_uJointFHThousandths[...]'s element as ULONG, but that array is
declared USHORT in data.c (confirmed pre-existing in the prior
PRUNE_VERIFY-removal commit, unrelated to that change).
Removed the entire #ifdef PRUNE_LIVE block in search.c (verified every
symbol inside it -- ShipDepthBucketPublic, JOINT_TABLE_SENTINEL_PUBLIC,
ShipLateBucketPublic, PruneVerifyPredict, PRUNE_LIVE_SENTINEL,
PruneLivePredict -- has no consumer anywhere outside that block) and
GNUmakefile's flag wiring, leaving an explanatory comment.
Verified: standard profile builds clean, precommit_check.sh passes.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Diffstat (limited to 'src/genetic/._main.cpp')
0 files changed, 0 insertions, 0 deletions
