retrieve .RData file of nameless object created with `magrittr::%>%`

I made a mistake and wrote this:

library(tidyverse)
list(x = rnorm(10)) %>% save(file = "test.RData")

instead of this:

x <- list(x = rnorm(10)) 
save(x, file = "test.RData")

As a result I cannot read "test.RData" back

load("test.RData")
ls()
character(0)

Yet the ".RData" file seems to contain something as the following command creates a 73 MB file compared with a 77 B with just ten values. It is just impossible to read back.

list(x = rnorm(10000000)) %>% save(file = "test.RData")

Interestingly the base r pipe |> throws an error

list(x = rnorm(10)) |> save(file = "test.RData")
Error in save(list(x = rnorm(10)), file = "test.RData") : 
  object ‘list(x = rnorm(10))’ not found

The mistake is easy to fix and I won't do it again but I am wondering what is happening under the hood:

Why is save() not throwing an error when passed a nameless object with %>%?
Would there be a way to load a .RData file written this way?

Solution

In fact it is loaded as the name . (dot) which was created by the %>% pipe. It is equivalent to the following which explicitly shows the name:

rnorm(10) %>% save(., file = "test.RData")

On the other hand the base pipe |> does not create a new name which accounts for the error if it is used. It works via substitution of the left hand side into the right hand side without creating a new name so is equivalent to the following which is an error:

save(rnorm(10), file = "test.RData") }

Note that

rnorm(10) |> save(file = "test.RData") |> quote()
## save(rnorm(10), file = "test.RData")

The reason the name does not appear is that ls does not show names that begin with dot unless called as

ls(all = TRUE)

Also note that load returns a character vector of names in it so that another way to get the name is to look at Names where

Names <- load("test.RData")
Names
## [1] "."