Make add_ggplot an S3 method

[schloerke]

• Situation:
  • GGally has ggplot2-like object "ggmatrix". I would like to be able to add themes and labels.
• Issue:
  • I currently do this by overwriting +.gg, which produces a warning.

library(ggplot2)
library(GGally)
#> Registered S3 method overwritten by 'GGally':
#>   method from   
#>   +.gg   ggplot2

^{Created on 2020-02-10 by the reprex package (v0.3.0)}

• Goal:
  • A way to customize how ggplot2-like objects and the 'right side objects' combine without producing the warning.

---

From https://r.789695.n4.nabble.com/quot-Incompatible-methods-quot-for-overloaded-operator-td4633362.html, Martin states that there can not be two + methods. 😞

from ?groupGeneric under 'Ops' (of which "+" is one)

       used.  If different methods are found, there is a warning
       about 'incompatible methods': in that case or if no method is
       found for either argument the internal method is used.

I have tried this with my ggmatrix class and can confirm it doesn't work as intended:

Error: (converted from warning) Incompatible methods ("+.ggmatrix", "+.gg") for "+"

I propose that that the add_ggplot method become an S3 method to allow for other classes to control the left side.

For my specific case, I would be able to add a method of add_ggplot.ggmatrix. From here, I can have a hook to the addition and not produce a warning.

cc @hafen

[schloerke]

If there is a better approach, I am all for it! Thank you!

[thomasp85]

ggplot_add() is already a generic which gets called by add_ggplot(). You can provide methods for that instead

[schloerke]

@thomasp85 While true, the current state still doesn't work.

ggplot_add is a generic for the right hand side of +.gg.

I'd like a generic for the left hand side of +.gg because I can't add any other + methods like +.ggmatrix.

---

Is there a way to define my own method if I know the classes of the right and left side?

From ?groupGeneric under Ops

The classes of both arguments are considered in dispatching
any member of this group. For each argument its vector of
classes is examined to see if there is a matching specific
(preferred) or ‘Ops’ method.

[yutannihilation]

We might consider integrating vctrs with ggplot2 (c.f. Hadley said so hadley/adv-r#1195 (comment)), which will enable double dispatch. Until then, maybe you can use S4 to do double dispatch?

[schloerke]

Pulling in @yutannihilation's example which matches my use case. My situation is exactly like p + theme_bw() at the bottom of the chunk.

library(ggplot2)

set.seed(100)
d1 <- data.frame(x = 1:100, y = cumsum(runif(100)))
d2 <- data.frame(x = 1:100, y = cumsum(runif(100)))

plot_all <- function(...) {
  l <- lapply(list(...), function(d) ggplot(d, aes(x, y)) + geom_line())
  l <- unname(l)
  class(l) <- "manyplot"
  l
}

print.manyplot <- function(x, ...) {
  do.call(gridExtra::grid.arrange, x)
}

p <- plot_all(d1, d2)
p

`+.manyplot` <- function(e1, e2) {
  l <- lapply(e1, function(x) x + e2)
  class(l) <- "manyplot"
  l
}

p + theme_bw()
#> Warning: Incompatible methods ("+.manyplot", "+.gg") for "+"
#> Error in p + theme_bw(): non-numeric argument to binary operator

---

maybe you can use S4?

Since both objects are not S4 objects, I believe Opts dispatch will not work.

https://stat.ethz.ch/R-manual/R-devel/library/methods/html/Methods_for_S3.html

However, primitive functions and operators are exceptions: The internal C code will look for S4 methods if and only if the object is an S4 object. S4 method dispatch would be used to dispatch any binary operator calls where either of the operands was an S4 object, for example.

Using an example from Advanced R, the S4 dispatch does not work as expected.

setGeneric("type", function(x) standardGeneric("type"))
#> [1] "type"
setMethod("type", signature("matrix"), function(x) "matrix")
setMethod("type", signature("character"), function(x) "character")

type(letters)
#> [1] "character"
type(matrix(letters, ncol = 2))
#> [1] "matrix"

foo <- structure(list(x = 1), class = "foo")
type(foo) # expected to fail
#> Error in (function (classes, fdef, mtable) : unable to find an inherited method for function 'type' for signature '"foo"'

setOldClass("foo")
setMethod("type", signature("foo"), function(x) "foo")
    
type(foo)
#> [1] "foo"
    
setMethod("+", signature(e1 = "foo", e2 = "numeric"), function(e1, e2) {
  structure(list(x = e1$x + e2), class = "foo")
})
foo + 3
#> Error in foo + 3: non-numeric argument to binary operator

^{Created on 2020-02-10 by the reprex package (v0.3.0)}

[yutannihilation]

I think I overcame this issue after this, but don't remember the details...

[yutannihilation]

I think I defined manyplot as a proper S4 class, not sure if this fits in your use case. Anyway, I don't think this is a good place to discuss here. Maybe we should continue on ggobi/ggally#292?

library(ggplot2)

set.seed(100)
d1 <- data.frame(x = 1:100, y = cumsum(runif(100)))
d2 <- data.frame(x = 1:100, y = cumsum(runif(100)))

setClass("manyplot", slots = c(plots = "list"))

plot_all <- function(...) {
  l <- lapply(list(...), function(d) ggplot(d, aes(x, y)) + geom_line())
  l <- unname(l)
  new("manyplot", plots = l)
}

setMethod("show", signature = "manyplot", function(object) {
  do.call(gridExtra::grid.arrange, object@plots)
})

p <- plot_all(d1, d2)
p

setOldClass(c("theme", "gg"))
setMethod("+", signature = c(e1 = "manyplot", e2 = "gg"), function(e1, e2) {
  e1@plots <- lapply(e1@plots, function(x) x + e2)
  e1
})

p + theme_bw()

^{Created on 2020-02-11 by the reprex package (v0.3.0)}