Dotplots with discrete y (#5463)

* Include y-offset

* Add test

* Add news bullet

Author: teunbrand

NEWS.md

@@ -1,5 +1,8 @@
 # ggplot2 (development version)
 
+* When using `geom_dotplot(binaxis = "x")` with a discrete y-variable, dots are
+  now stacked from the y-position rather than from 0 (@teunbrand, #5462)
+
 * (breaking) In the `scale_{colour/fill}_gradient2()` and 
   `scale_{colour/fill}_steps2()` functions, the `midpoint` argument is 
   transformed by the scale transformation (#3198).

R/geom-dotplot.R

@@ -242,12 +242,12 @@ GeomDotplot <- ggproto("GeomDotplot", Geom,
       # ymin, ymax, xmin, and xmax define the bounding rectangle for each stack
       # Can't do bounding box per dot, because y position isn't real.
       # After position code is rewritten, each dot should have its own bounding box.
+      yoffset <- if (is_mapped_discrete(data$y)) data$y else 0
       data$xmin <- data$x - data$binwidth / 2
       data$xmax <- data$x + data$binwidth / 2
-      data$ymin <- stackaxismin
-      data$ymax <- stackaxismax
-      data$y    <- 0
-
+      data$ymin <- stackaxismin + yoffset
+      data$ymax <- stackaxismax + yoffset
+      data$y <- yoffset
     } else if (params$binaxis == "y") {
       # ymin, ymax, xmin, and xmax define the bounding rectangle for each stack
       # Can't do bounding box per dot, because x position isn't real.

tests/testthat/_snaps/geom-dotplot/bin-x-three-y-groups-stack-centerwhole.svg

@@ -154,6 +154,9 @@ test_that("geom_dotplot draws correctly", {
   # Binning along y, with multiple grouping factors
   dat2 <- data_frame(x = rep(factor(LETTERS[1:3]), 30), y = rnorm(90), g = rep(factor(LETTERS[1:2]), 45))
 
+  expect_doppelganger("bin x, three y groups, stack centerwhole",
+    ggplot(dat2, aes(y, x)) + geom_dotplot(binwidth = .25, binaxis = "x", stackdir = "centerwhole")
+  )
   expect_doppelganger("bin y, three x groups, stack centerwhole",
     ggplot(dat2, aes(x, y)) + geom_dotplot(binwidth = .25, binaxis = "y", stackdir = "centerwhole")
   )