use ‘{{}}’

make_scatter_plot1<-function(colorVar) {
  plt<-ggplot(mtcars, aes(x=wt, y=mpg, color=factor({{colorVar}}))) + theme_bw()
  plt<-plt+geom_point()
  plt
}
make_scatter_plot1(cyl)

The embrace operator ‘⁠{{’⁠ is used to create functions that call other data-masking functions. It transports a data-masked argument (an argument that can refer to columns of a data frame) from one function to another.

use .data

make_scatter_plot2<-function(colorVar) {
  plt<-ggplot(mtcars, aes(x=wt, y=mpg, color=factor(.data[[colorVar]]))) + theme_bw()
  plt<-plt+geom_point()
  plt
}
make_scatter_plot2("cyl")

use sym()

make_scatter_plot3<-function(colorVar) {
  plt<-ggplot(mtcars, aes(x=wt, y=mpg, color=factor(!!sym(colorVar)) )) + theme_bw()
  plt<-plt+geom_point()
  plt
}
make_scatter_plot3("cyl")

use quo()

make_scatter_plot4<-function(colorVar) {
  plt<-ggplot(mtcars, aes(x=wt, y=mpg, color=factor(!!colorVar) )) + theme_bw()
  plt<-plt+geom_point()
  plt
}
myvar<-quo(cyl)
make_scatter_plot4(myvar)

defusing

When a piece of R code is defused, R doesn’t return its value, but returns the expression in a tree-like object to describe how to compute a value, which can be evaluated later in a specific environment.

For example, one can use expr() to create a defused object.

# return result
1+1
## [1] 2
# return defused object
expr(1+1)
## 1 + 1

One application of using defused object is to resume its evaluation in data mask, see the example below:

e <- expr(mean(cyl))
eval(e, mtcars)
## [1] 6.1875

Some functions defuse their arguments, and when these function are called from another function, its argument must be provided with embrace operator {{, which is a combination of enquo() and !!.

Defused expressions can also be created by using other functions such as sym() and call(), for example:

# create a symbol
vName<-"mtcars"
sym(vName)

# make a call
call("head", sym(vName))

There are two ways to defuse expressions, one can use expr() to defuse one’s own expression and use enquo() to defuse user-provided expressions to your function.

Defusing is known as quoting in other frameworks.

• The equivalent of expr() is [base::bquote()].

• The equivalent of enquo() is [base::substitute()]. The latter returns a naked expression instead of a quosure.

• There is no equivalent for enquos(…) but you can defuse dots as a list of naked expressions with eval(substitute(alist(…))).

data-masking

Data masking means that a data.frame is inserted at the bottom of the current chain of environment and it takes higher precedence than the user environment.

For example, if we assign a value to a variable, but this variable is invisible in data-masking functions as below:

mpg<-100
# the following function can't see it
dplyr::summarise(mtcars, mean(mpg))
##   mean(mpg)
## 1  20.09062

As you can see, the function used the column mpg from the dataframe mtcars.

One way to disambiguate the user and mask environments is to use tidy eval pronouns.

dplyr::summarise(mtcars, mean(.env$mpg))
##   mean(.env$mpg)
## 1            100