Coord sf breaks fallback

[teunbrand]

This PR aims to fix #2985.

Briefly, if the graticule lines are horizontal or vertical, we could infer where the tickmark should be placed. This PR doesn't solve axis labels placement for all projections, but it does fix a common painpoint.

For WGS84, it solves the problem:

devtools::load_all("~/packages/ggplot2/")
#> ℹ Loading ggplot2

world <- sf::st_as_sf(rworldmap::getMap(resolution = "low"))
p <- ggplot(world) + 
  geom_sf()
p

For the Robinson projection, it can place the y-axis labels and a single label at 0 degrees.

p + coord_sf(crs = "+proj=robin")

For this projection, no labels can be placed because all graticule lines are curved.

p + coord_sf(crs = "+proj=laea +lat_0=52 +lon_0=10 +x_0=4321000 +y_0=3210000")

The IGH projection only gets y-axis labels because all x-graticule lines are not straight.

p + coord_sf(crs = "+proj=igh")

^{Created on 2024-12-11 with reprex v2.1.1}

[thomasp85]

LGTM, though the situation with axes and sf continues to bring me pain

[teunbrand]

the situation with axes and sf continues to bring me pain

Me too, I wish there was an sf-based solution that would just work for all types of projections

[edzer]

Looks great, thanks - I'll try to dig into this and see how we can be more helpful from the side of sf in this respect (and other packages plotting maps with R).

[teunbrand]

That would be amazing to have @edzer! When I was trying to implement in-panel placement of axes, the problem I kept running into was not too much the ticks/label placement, but the axis line placement. We'd have to know the outline/extent of a map to gracefully place an axis. If that were easy to get from {sf} for most projections that'd really make a big dent in the larger issue.