Not all items in the list aes() produces are quosures

[rpruim]

https://github.com/tidyverse/ggplot2/blob/master/NEWS.md#ggplot2-2219000 claims

The primary developer facing change is that aes() is now a list of quosures (expression + environment pairs) rather than a list of symbols, and you'll need to take a different approach to extracting the information you need.

But not all elements are quosures:

aes(x = foo, y = 0) %>% str()
## List of 2
##  $ x: language ~foo
##   ..- attr(*, ".Environment")=<environment: R_GlobalEnv> 
##  $ y: num 0
##  - attr(*, "class")= chr "uneval"
aes(x = foo, y = 0) %>% lapply(class)
## $x
## [1] "quosure" "formula"
## 
## $y
## [1] "numeric"
## 
aes(x = foo, y = 0) %>% lapply(function(x) inherits(x, "quosure"))
## $x
## [1] TRUE
## 
## $y
## [1] FALSE

[yutannihilation]

It seems this is the intended behaviour as the code introduced by 1f39e13 says:

ggplot2/R/aes.r

Lines 100 to 101 in 0c7eda0

	# Wrap symbolic objects in quosures but pull out constants out of
	# quosures for backward-compatibility

This seems somehow consistent with the behaviour of the CRAN version where constants are not quoted (because simply they cannot be quoted (e.g. quote(1) returns 1)):

aes(x = foo, y = 0) %>% str()
#> List of 2
#>  $ x: symbol foo
#>  $ y: num 0

That said, I too feel the NEWS may be misleading a bit...

[lock]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/