Opened 6 years ago
Last modified 3 years ago
#5092 closed defect
ClumpPlacer seems to use incorrect formula to determine if placement failed for a given failFraction allowed. — at Version 1
| Reported by: | Inari | Owned by: | |
|---|---|---|---|
| Priority: | Should Have | Milestone: | Alpha 26 |
| Component: | Maps | Keywords: | |
| Cc: | Patch: |
Description (last modified by )
To check if placement has failed, ClumpPlacer allows to pass a failFraction to allow partial placement to still count as success.
(ps/trunk/binaries/data/mods/public/maps/random/rmgen/placer/centered/ClumpPlacer.js#L103)
The formula to check if placement has failed seems a bit off though. It keeps track of a failed variable, increasing that by 1 each time a point fails to put.
In a quick test for an area with size 1200 there seem to be roughly 13500~15500 points that it attempts to put. It varies a bit because of the randomness involved in the placement.
So it can't just check that failed variable against the size multiplied by failFraction. It should instead keep track of how many points it tried to put and multiply that count by failFraction to check against failed.
ChainPlacer seems to do this already (ps/trunk/binaries/data/mods/public/maps/random/rmgen/placer/centered/ChainPlacer.js#L110).
This is also why #4810 seemed to be an issue, originally it passed a failFraction of 1, allowing for placement to proceed in any case. But that seems to not have worked as intended.
