Closed dfellis closed 5 years ago
Should not have doubted the h3-js
implementation. It is the correct one. Need to figure out what I've done wrong with the other.
Fixed, now they match. There were three classes of issues -- overloading the i
variable instead of using i
and j
, accidentally setting lng
on the main geofence when setting the hole lng
, and accidentally setting the hole data in a local copy of the hole instead of on the calloc
ed hole record.
I think making the polyfill
binding function easier to read can happen in another PR, along with general cleanup of the macros so they're more composable than they currently are.
This is not 100% done. I need to DRY up the code a bit, which currently consists of a lot of copy-pasted logic, and I need to figure out what's going on with the polyfill with holes test. It's not matching the h3-js results, but I suspect this may be a bug in h3-js and not h3-node?