How to create a vector between pipes as intermediate step?

Question

Within a piping chain, I am trying to store the variable names of the current data frame/tibble, and then continue.

In the example below, how do we get dplyr::filter() to receive the iris object? tmp should be used further down the chain. This attempt fails:

library(dplyr) 
iris %>% 
  {names(select(., where(is.numeric))) ->> tmp} %>%
  filter(., Species=="setosa")

Answer 1

You can just slightly modify your existing code by adding a ...; .} at the end of the intermediate line:

iris %>% 
  {names(select(., where(is.numeric))) ->> tmp; .} %>%
  filter(., Species=="setosa")

Though it may be more efficient to define it out of the pipe (i.e. tmp <- names(select(iris, where(is.numeric))).

Answer 2

If the only reason to create tmp is to use it later in the pipeline then, in general, it is preferable to keep everything local within the pipeline rather than use a global variable. We can do that by assigning tmp locally rather than globally and then, along with the definition of tmp, placing the entire remainder of the pipeline within the {...} .

library(dplyr) 

iris %>% 
  { tmp <- names(select(., where(is.numeric)))
    filter(., Species=="setosa") %>%
    select(-Species) %>% 
    matplot(main = toString(tmp))
  }

(continued after graph)

Of course we could alternately break this up into two pipelines:

tmp <- iris %>%
  select(where(is.numeric)) %>%
  names

iris %>%
  filter(Species=="setosa") %>%
  select(-Species) %>% 
  matplot(main = toString(tmp))