Release 4.0.3

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: ggplot2
 Title: Create Elegant Data Visualisations Using the Grammar of Graphics
-Version: 4.0.0.9000
+Version: 4.0.2.9000
 Authors@R: c(
     person("Hadley", "Wickham", , "[email protected]", role = "aut",
            comment = c(ORCID = "0000-0003-4757-117X")),

NEWS.md

@@ -1,25 +1,39 @@
-# ggplot2 (development version)
+# Development version
+
+This is a hotfix release fixing a test, a bug and introducing a feature.
+
+* Adapt to changes in `rlang::warn_dots_used()` (@lionel-, #6830).
+* Fixed bug where `guide_axis_theta()` didn't anticipate old-style text elements (#6803).
+* `geom_boxplot()`/`stat_boxplot()` gain a `quantile.type` parameter (default `7`) 
+   to control the percentile definition used for hinges and median; set `quantile.type = 2` 
+   to match SAS's default `PCTLDEF = 5`, enabling parity with SAS boxplots out-of-the-box. 
+   (@munoztd0, #6819)
+
+# ggplot2 4.0.2
+
+This hotfix release makes ggplot2 more interoperable between rlang versions.
+
+* `make_constructor()` no longer captures `rlang::list2()` at build time.
+
+# ggplot2 4.0.1
+
+This is a smaller patch release focussed on fixing regressions from 4.0.0 and 
+polishing the recent features.
+
+## Bug fixes
 
 * Fixed regression where `geom_area()` didn't draw panels with single groups 
   when `stat = "align"` (@teunbrand, #6680)
 * Fixed regression where `position_stack(vjust)` was ignored when there are
   only single groups (#6692)
 * Fixed bug where `NA` handling in `geom_path()` was ignoring panels (@teunbrand, #6533)
-* Logical values for the linetype aesthetic will be interpreted numerically,
-  so that `linetype = FALSE` becomes 0/'blank' and `linetype = TRUE` becomes 
-  1/'solid' (@teunbrand, #6641)
-* Out-of-bounds datapoints used as padding by `stat_align()` now get removed
-  silently rather than verbosely (@teunbrand, #6667)
 * Fixed bug where `stat_bin(boundary)` was ignored (#6682).
 * `geom_text()` and `geom_label()` accept expressions as the `label` aesthetic 
   (@teunbrand, #6638)
 * Fixed regression where `draw_key_rect()` stopped using `fill` colours 
   (@mitchelloharawild, #6609).
 * Fixed regression where `scale_{x,y}_*()` threw an error when an expression
   object is set to `labels` argument (@yutannihilation, #6617).
-* Improved palette fallback mechanism in scales (@teunbrand, #6669).
-* Allow `stat` in `geom_hline`, `geom_vline`, and `geom_abline`. (@sierrajohnson, #6559)
-* `stat_boxplot()` treats `width` as an optional aesthetic (@Yunuuuu, #6575)
 * Fixed regression where the first (unnamed) argument to colour/fill scales was 
   not passed as the `name` argument (@teunbrand, #6623)
 * Fixed issue where vectorised `arrow()`s caused errors in drawing the 
@@ -28,10 +42,21 @@
   insistently. Now they contribute only as fallback labels (@teunbrand, #6616)
 * Fixed regression where empty arguments to colour/fill scale caused errors
   (@jmbarbone, #6710)
+* Fixed axis misplacement in `coor_radial()` when labels are blank (@teunbrand, #6574)
+
+## Improvements
+
+* Improved palette fallback mechanism in scales (@teunbrand, #6669).
+* Allow `stat` in `geom_hline`, `geom_vline`, and `geom_abline`. (@sierrajohnson, #6559)
+* `stat_boxplot()` treats `width` as an optional aesthetic (@Yunuuuu, #6575)
 * The `theme(panel.widths, panel.heights)` setting attempts to preserve the
   plot's aspect ratio when only one of the two settings is given, and the plot 
   has a single panel (@teunbrand, #6701).
-* Fixed axis misplacement in `coor_radial()` when labels are blank (@teunbrand, #6574)
+* Logical values for the linetype aesthetic will be interpreted numerically,
+  so that `linetype = FALSE` becomes 0/'blank' and `linetype = TRUE` becomes 
+  1/'solid' (@teunbrand, #6641)
+* Out-of-bounds datapoints used as padding by `stat_align()` now get removed
+  silently rather than verbosely (@teunbrand, #6667)
 
 # ggplot2 4.0.0
 
@@ -125,6 +150,8 @@
     * `header_family` to easily set the font for headers and titles (#5886)
         * To accommodate, `plot.subtitle`, `plot.caption` and `plot.tag` now
           inherit from the root `text` element instead of the `title` element.
+    * The `base_size` argument also propagates to settings in the `theme(geom)`
+      setting.
 * New function family for setting parts of a theme. For example, you can now use
   `theme_sub_axis(line, text, ticks, ticks.length, line)` as a substitute for
   `theme(axis.line, axis.text, axis.ticks, axis.ticks.length, axis.line)`. This

R/facet-.R

@@ -1421,14 +1421,14 @@ censor_labels <- function(ranges, layout, labels) {
   )
 
   if (!labels$x) {
-    xmax <- stats::ave(layout$ROW, layout$COL, FUN = max)
-    xmin <- stats::ave(layout$ROW, layout$COL, FUN = min)
+    xmax <- vec_ave(layout$ROW, layout$COL, max)
+    xmin <- vec_ave(layout$ROW, layout$COL, min)
     draw[which(layout$ROW != xmax), "bottom"] <- FALSE
     draw[which(layout$ROW != xmin), "top"] <- FALSE
   }
   if (!labels$y) {
-    ymax <- stats::ave(layout$COL, layout$ROW, FUN = max)
-    ymin <- stats::ave(layout$COL, layout$ROW, FUN = min)
+    ymax <- vec_ave(layout$COL, layout$ROW, max)
+    ymin <- vec_ave(layout$COL, layout$ROW, min)
     draw[which(layout$COL != ymax), "right"] <- FALSE
     draw[which(layout$COL != ymin), "left"] <- FALSE
   }

R/geom-path.R

@@ -19,7 +19,7 @@ GeomPath <- ggproto("GeomPath", Geom,
     # middle since you expect those to be shown by a break in the line
     aesthetics <- c(self$required_aes, self$non_missing_aes)
     complete <- stats::complete.cases(data[names(data) %in% aesthetics])
-    kept <- stats::ave(complete, data$group, data$PANEL, FUN = keep_mid_true)
+    kept <- vec_ave(complete, interaction(data$group, data$PANEL, drop = TRUE), keep_mid_true)
     data <- data[kept, ]
 
     if (!all(kept) && !params$na.rm) {
@@ -48,7 +48,7 @@ GeomPath <- ggproto("GeomPath", Geom,
     munched <- coord_munch(coord, data, panel_params)
 
     # Silently drop lines with less than two points, preserving order
-    rows <- stats::ave(seq_len(nrow(munched)), munched$group, FUN = length)
+    rows <- vec_ave(seq_len(nrow(munched)), munched$group, length)
     munched <- munched[rows >= 2, ]
     if (nrow(munched) < 2) return(zeroGrob())
 

R/guide-.R

@@ -284,7 +284,7 @@ Guide <- ggproto(
 
     if (is.numeric(breaks)) {
       range <- scale$continuous_range %||% scale$get_limits()
-      key <- vec_slice(key, is.finite(oob_censor_any(breaks, range)))
+      key <- key[is.finite(oob_censor_any(breaks, range)), , drop = FALSE]
     } else {
       key
     }

R/guide-axis-theta.R

@@ -198,7 +198,11 @@ GuideAxisTheta <- ggproto(
 
     # Resolve text angle
     if (is_waiver(params$angle) || is.null(params$angle)) {
-      angle <- elements$text@angle
+      angle <- if (S7::S7_inherits(elements$text)) {
+        elements$text@angle
+      } else {
+        elements$text$angle
+      }
     } else {
       angle <- flip_text_angle(params$angle - rad2deg(key$theta))
     }

R/guide-bins.R

@@ -181,7 +181,7 @@ GuideBins <- ggproto(
     }
 
     key$.label <- labels
-    key <- vec_slice(key, !is.na(oob_censor_any(key$.value)))
+    key <- key[!is.na(oob_censor_any(key$.value)), , drop = FALSE]
 
     return(key)
   },

R/layer.R

@@ -972,10 +972,8 @@ normalise_label <- function(label) {
     return(NULL)
   }
   if (obj_is_list(label)) {
-    # Ensure that each element in the list has length 1
+    # Ensure no elements are empty
     label[lengths(label) == 0] <- ""
-    truncate <- !vapply(label, is.call, logical(1)) # Don't mess with call/formula
-    label[truncate] <- lapply(label[truncate], `[`, 1)
   }
   if (is.expression(label)) {
     # Classed expressions, when converted to lists, retain their class.

R/make-constructor.R

@@ -140,7 +140,7 @@ make_constructor.Geom <- function(x, ..., checks = exprs(), omit = character(),
   body <- call2("{", !!!checks, body)
 
   # We encapsulate rlang::list2
-  new_env <- new_environment(list(list2 = list2), env)
+  new_env <- new_environment(list(list2 = list2_wrapper), env)
 
   new_function(fmls, body, new_env)
 }
@@ -225,8 +225,11 @@ make_constructor.Stat <- function(x, ..., checks = exprs(), omit = character(),
   body <- call2("{", !!!checks, body)
 
   # We encapsulate rlang::list2
-  new_env <- new_environment(list(list2 = list2), env)
+  new_env <- new_environment(list(list2 = list2_wrapper), env)
 
   new_function(fmls, body, new_env)
 }
 
+list2_wrapper = function(...) {
+  rlang::list2(...)
+}

R/position-dodge.R

@@ -150,13 +150,13 @@ PositionDodge <- ggproto("PositionDodge", Position,
     }
 
     data$order <- xtfrm( # xtfrm makes anything 'sortable'
-      data$order %||% ave(data$group, data$x, data$PANEL, FUN = match_sorted)
+      data$order %||% vec_ave(data$group, data[c("x", "PANEL")], fn = match_sorted)
     )
     if (isTRUE(params$reverse)) {
       data$order <- -data$order
     }
     if (is.null(params$n)) { # preserve = "total"
-      data$order <- ave(data$order, data$x, data$PANEL, FUN = match_sorted)
+      data$order <- vec_ave(data$order, data[c("x", "PANEL")], fn = match_sorted)
     } else { # preserve = "single"
       data$order <- match_sorted(data$order)
     }

R/stat-boxplot.R

@@ -53,15 +53,16 @@ StatBoxplot <- ggproto("StatBoxplot", Stat,
 
   extra_params = c("na.rm", "orientation"),
 
-  compute_group = function(data, scales, width = NULL, na.rm = FALSE, coef = 1.5, flipped_aes = FALSE) {
+  compute_group = function(data, scales, width = NULL, na.rm = FALSE, coef = 1.5, flipped_aes = FALSE, quantile.type = 7) {
     data <- flip_data(data, flipped_aes)
     qs <- c(0, 0.25, 0.5, 0.75, 1)
 
     if (!is.null(data$weight)) {
       mod <- quantreg::rq(y ~ 1, weights = weight, data = data, tau = qs)
       stats <- as.numeric(stats::coef(mod))
     } else {
-      stats <- as.numeric(stats::quantile(data$y, qs))
+      # Follow base R default (type = 7) unless overridden by user
+      stats <- as.numeric(stats::quantile(data$y, qs, type = quantile.type))
     }
     names(stats) <- c("ymin", "lower", "middle", "upper", "ymax")
     iqr <- diff(stats[c(2, 4)])
@@ -99,6 +100,8 @@ StatBoxplot <- ggproto("StatBoxplot", Stat,
 
 #' @rdname geom_boxplot
 #' @param coef Length of the whiskers as multiple of IQR. Defaults to 1.5.
+#' @param quantile.type An integer between 1 and 9 setting the quantile algorithm
+#'   per [`stats::quantile(type)`][stats::quantile]. Defaults to `7`
 #' @inheritParams stat_identity
 #' @export
 #' @eval rd_computed_vars(

R/stat-sum.R

@@ -15,7 +15,7 @@ StatSum <- ggproto(
 
     counts <- count(data, group_by, wt_var = "weight")
     counts <- rename(counts, c(freq = "n"))
-    counts$prop <- stats::ave(counts$n, counts$group, FUN = prop.table)
+    counts$prop <- vec_ave(counts$n, counts$group, prop.table)
     counts
   }
 )

R/stat-ydensity.R

@@ -73,7 +73,7 @@ StatYdensity <- ggproto(
           "{.arg quantiles} for weighted data is not implemented."
         )
       }
-      quants <- quantile(data$y, probs = quantiles)
+      quants <- stats::quantile(data$y, probs = quantiles)
       quants <- data_frame0(
         y = unname(quants),
         quantile = quantiles
@@ -82,7 +82,7 @@ StatYdensity <- ggproto(
       # Interpolate other metrics
       for (var in setdiff(names(dens), names(quants))) {
         quants[[var]] <-
-          approx(dens$y, dens[[var]], xout = quants$y, ties = "ordered")$y
+          stats::approx(dens$y, dens[[var]], xout = quants$y, ties = "ordered")$y
       }
 
       dens <- vec_slice(dens, !dens$y %in% quants$y)

R/utilities.R

@@ -910,7 +910,7 @@ compute_data_size <- function(data, size, default = 0.9,
     res <- vapply(res, resolution, FUN.VALUE = numeric(1), ...)
     res <- min(res, na.rm = TRUE)
   } else if (panels == "by") {
-    res <- stats::ave(data[[var]], data$PANEL, FUN = function(x) resolution(x, ...))
+    res <- vec_ave(data[[var]], data$PANEL, function(x) resolution(x, ...))
   } else {
     res <- resolution(data[[var]], ...)
   }
@@ -930,3 +930,14 @@ try_prop <- function(object, name, default = NULL) {
   }
   S7::prop(object, name)
 }
+
+vec_ave <- function(x, by, fn, ...) {
+  idx <- vec_group_loc(by)$loc
+  list_unchop(
+    lapply(
+      vec_chop(x, indices = idx),
+      FUN = fn, ...
+    ),
+    indices = idx
+  )
+}

README.Rmd

@@ -76,7 +76,7 @@ If you are new to ggplot2 you are better off starting with a systematic introduc
     [Communication][r4ds-comm] chapters in
     [R for Data Science][r4ds]. R for Data Science is designed to
     give you a comprehensive introduction to the
-    [tidyverse](https://www.tidyverse.org), and these two chapters will
+    [tidyverse](https://tidyverse.org/), and these two chapters will
     get you up to speed with the essentials of ggplot2 as quickly as
     possible.
 
@@ -119,4 +119,4 @@ There are two main places to get help with ggplot2:
 [r4ds-vis]: https://r4ds.hadley.nz/data-visualize
 [r4ds-comm]: https://r4ds.hadley.nz/communication
 [oreilly]: https://learning.oreilly.com/videos/data-visualization-in/9781491963661/
-[blog]: https://www.tidyverse.org/tags/ggplot2/
+[blog]: https://tidyverse.org/tags/ggplot2/

README.md

@@ -80,8 +80,8 @@ documentation pages. Currently, there are several good places to start:
     [Communication](https://r4ds.hadley.nz/communication) chapters in [R
     for Data Science](https://r4ds.hadley.nz). R for Data Science is
     designed to give you a comprehensive introduction to the
-    [tidyverse](https://www.tidyverse.org), and these two chapters will
-    get you up to speed with the essentials of ggplot2 as quickly as
+    [tidyverse](https://tidyverse.org/), and these two chapters will get
+    you up to speed with the essentials of ggplot2 as quickly as
     possible.
 
 2.  If you’d like to take an online course, try [Data Visualization in R
@@ -105,7 +105,7 @@ documentation pages. Currently, there are several good places to start:
     graphics specifically tailored to your needs.
 
 6.  For articles about announcements and deep-dives you can visit the
-    [tidyverse blog](https://www.tidyverse.org/tags/ggplot2/).
+    [tidyverse blog](https://tidyverse.org/tags/ggplot2/).
 
 ## Getting help
 

_pkgdown.yml

@@ -18,7 +18,8 @@ template:
       data-project-name="ggplot2"
       data-modal-title="tidyverse AI ✨"
       data-project-logo="https://avatars.githubusercontent.com/u/22032646?s=200&u=235532df5cf8543246812f73db051b793f868807&v=4"
-      data-user-analytics-fingerprint-enabled="true"
+      data-user-analytics-fingerprint-enabled="false"
+      data-user-analytics-cookie-enabled="false"
       data-bot-protection-mechanism="hcaptcha"
       data-website-id="5dcac24a-bd5a-4874-9f5b-3341afddd8c3"
       ></script>

cran-comments.md

@@ -1,9 +1,2 @@
-This is a major release of ggplot2. Among other things it includes an adaption
-of S7 for all the S3 classes and methods that were currently in use.
-Unfortunately such a change is not without issue with the number of reverse
-dependencies that ggplot2 has as many packages wrongfully checks the internals
-of ggplot2 objects in their tests.
-
-Because of this you should expect a larger than usual number of breaking
-packages. We have been very diligent to reach out to all maintainers over the
-last 2-3 months and provided resolutions but breakages are still to be expected.
+This is a patch release that resolves a failing test and supports some minimal bug fixes and a small uninvasive feature.
+We expect no problems with reverse dependencies.