search

ocean-tracking-network / glatos

9 stars 4 forks source link

Remove `geosphere` dependency #236

Closed mhpob closed 3 months ago

mhpob commented 6 months ago

Current R CMD check failure is due to a floating point issue on macOS which will be fixed with acceptance of https://github.com/ocean-tracking-network/glatos/pull/228

This PR removes dependency on geosphere (See #235).

May introduce a breaking change in point_offset by using the Haversine (great circle) method as originally described in the documentation as opposed to a geodesic method (ellipsoid) as used by geosphere. Creates differences on the order of <5 meters in distances less than a few dozen kilometers, but can increase to nearly 2 kilometers on the scale of the entire Great Lakes area.

Adds initial tests for point_offset and interpolate_path as none currently exist.

chrisholbrook commented 3 months ago

Given that this function was really just a wrapper for geoshere::destPoint() that allows input in the form of compass point abbreviations (e.g., "NNE", "S"), I support this change and also suggest we deprecate this function in favor of a new function that simply converts the compass abbreviations to (nominal) numerical equivalents. Then suggest using geosphere::distPoint() with those as input.

point_offset() is not used anywhere within this package, though I am aware of at least two people who use it to derive approximate location of snag line anchors (to facilitate receiver recovery).