Inconsistent select() and rename() behavior

[jmobrien]

Passing a named list to select() performs select/rename behavior, but doing the same with rename() produces the confusing error "All arguments must be named"

It does work when "!!!" is applied, but I'm unclear what's different about the structure here, and the message doesn't really provide insight.

newnames <- c("Sepal.Length", "Petal.Length") %>% 
  setNames(c("Sepal_Length", "Petal_Length"))

iris %>% select(newnames) %>% names
iris %>% rename(newnames) %>% names

iris %>% select(!!!newnames) %>% names
iris %>% rename(!!!newnames) %>% names

[batpigandme]

From the rename() docs

Use named arguments, e.g. new_name = old_name, to rename selected variables.
To turn the named character vector into named arguments, you have to use the splice !!! (big bang) operator.

Hopefully we can make the error message more informative, though.

library(dplyr)
newnames <- c("Sepal.Length", "Petal.Length") %>% 
  setNames(c("Sepal_Length", "Petal_Length"))

iris %>% select(newnames) %>% names
#> [1] "Sepal_Length" "Petal_Length"

iris %>% rename(newnames) %>% names
#> All arguments must be named

iris %>% select(!!!newnames) %>% names
#> [1] "Sepal_Length" "Petal_Length"
iris %>% rename(!!!newnames) %>% names
#> [1] "Sepal_Length" "Sepal.Width"  "Petal_Length" "Petal.Width" 
#> [5] "Species"

vars <- list(
  Sepal_Length = sym("Sepal.Length"),
  Petal_Length = sym("Petal.Length")
)

rename(iris, !!!vars) %>% head()
#>   Sepal_Length Sepal.Width Petal_Length Petal.Width Species
#> 1          5.1         3.5          1.4         0.2  setosa
#> 2          4.9         3.0          1.4         0.2  setosa
#> 3          4.7         3.2          1.3         0.2  setosa
#> 4          4.6         3.1          1.5         0.2  setosa
#> 5          5.0         3.6          1.4         0.2  setosa
#> 6          5.4         3.9          1.7         0.4  setosa

^{Created on 2019-04-09 by the reprex package (v0.2.1)}

EDIT For completeness, it also works without sym() with the !! operator:

library(dplyr)
vars <- c(Sepal_Length = "Sepal.Length", Petal_Length ="Petal.Length")
iris %>% select(!!vars) %>% names
#> [1] "Sepal_Length" "Petal_Length"
iris %>% rename(!!vars) %>% names
#> [1] "Sepal_Length" "Sepal.Width"  "Petal_Length" "Petal.Width" 
#> [5] "Species"

^{Created on 2019-04-09 by the reprex package (v0.2.1)}

[jmobrien]

Thanks--though the text you copied is not actually in the help docs? This line is not followed by what you copied:

Use named arguments, e.g. new_name = old_name, to rename selected variables.

Am I missing something?

Also, it's more the inconsistency that troubles me--why does it work with select, then? The two do share a help page, after all, and I thought they largely shared functionality.

[batpigandme]

ping @lionel- , is this a tidyselect issue?

[lionel-]

Yes most likely.

[lionel-]

We're going to deprecate this pattern:

iris %>% select(newnames)

The correct way to do it will be one of these (first one is recommended):

iris %>% select(one_of(newnames))
iris %>% select(!!newnames)

The reason is that if the data frame contains a newnames column, you'll get unwanted results. Conversely if you do select(x) because you expect a column named x, but there isn't one, and instead a variable x is picked up in the environment containing integers or names, you'll get surprising results.

For these reasons, we have decided to make a clear delimitation between expressions referring to columns and expressions referring to the context, like one_of(). As usual, unquoting with !! is another way of evaluating in the context unambiguously. Note that these considerations only apply to selections, not to actions (as in mutate(), summarise(), etc). See #76 for more about this.

The bottom line is that you need to unquote the character vector:

iris %>% rename(!!newnames)