Closed whuzs closed 1 year ago
That was/is an experimental traversal algorithm that fixed quite a few performance degenration issues for situations where many of the queries happen far away from the actual data points - for which the regular k-d tree traversal could suffer quite a bit. I've completely rewritten that in a different branch, so it's possible that the one in main is currently broken. I'll have a look
can you tell me for what exactly it fails verification? any reproducer would help
oh, wait, i remember how/when i fixed it in that other branch .... the bug isn't actually in the traversal, it's in the float atomic_min/max for computing the bouding box. if the initial bounding box is wrong then traversal misses some branches. i'll fix this soon - just trying to do it with a minimum of changes so i don't end up in merge conflict purgatory later on - all that code got completely rewritten in that other branch. one sec.
Thank you for your reply. Yes, I verified an error during my previous testing of testCaseID2, but it has now been fixed.
What does this function do? i found enable this function but some case verify failed.