Dan Kelley

Hm `sandbox/dk/eow/eow02.R` with a lat_0 gives odd results. These seem independent of the number of steps in the bisection search. Next, I'll try using a grid pattern, as a way...

Hm, The weird outline is not a result of bisection, as the grid test shows. I think `map2lonlat()` must be doing something odd. Note the interesting thing at the bottom-centre...

Focussing in on one spot, and doing direct calls to the `sf` function (to remove possibility that the hiccough is caused by an `oce` wrapper), I see that these `x,y`...

A test case: -75.612 N and -85.352 E (using google maps to find the spot. I like their new interface, but it won't let you centre on either pole ......

I made a test that is independent of `oce`, as shown in the reprex pasted below. Red dots in the diagram indicate that the projected points can be reverse-projected. By...

Hm, `rgdal` can do this. (I've not checked yet on whether it reliably gives `NA`s outside the earth ... that's what I need.) ``` r library(rgdal) #> Loading required package:...

Any chance of adding a `sessionInfo()` note? (PS. I submitted an issue to r-spatial/sf and perhaps someone there will want to know this info.)

@clayton33 can you do `sf::sf_extSoftVersion()` and post the results at https://github.com/r-spatial/sf/issues/1729? That might help in the analysis of how things are different. Many thanks. Dan.

Wow, @richardsc is getting *very* different results. I will be summarizing all the results in a text file, for transmission to the "sf" team. Thanks!

I've put notes in the `sandbox/dk/eow/README.md` file.