In this lab, we will learn how to get information about environments and the search path.
Goal: by the end of this lab, you should be able to understand how loading packages affects the search path.
Understanding the search path is crucial to understanding how R looks for the values that are bound to names. When you start a new R session, the search path contains only those packages that are loaded by default.
search()## [1] ".GlobalEnv" "package:htmltools" "package:knitr"
## [4] "package:forcats" "package:stringr" "package:dplyr"
## [7] "package:purrr" "package:readr" "package:tidyr"
## [10] "package:tibble" "package:ggplot2" "package:tidyverse"
## [13] "package:stats" "package:graphics" "package:grDevices"
## [16] "package:utils" "package:datasets" "package:methods"
## [19] "Autoloads" "package:base"Note that if we use a function from another package using the
:: operator, the package is loaded, but it
is not added to the search path (i.e., attached).
rlang::search_envs()## [[1]] $ <env: global>
## [[2]] $ <env: package:htmltools>
## [[3]] $ <env: package:knitr>
## [[4]] $ <env: package:forcats>
## [[5]] $ <env: package:stringr>
## [[6]] $ <env: package:dplyr>
## [[7]] $ <env: package:purrr>
## [[8]] $ <env: package:readr>
## [[9]] $ <env: package:tidyr>
## [[10]] $ <env: package:tibble>
## [[11]] $ <env: package:ggplot2>
## [[12]] $ <env: package:tidyverse>
## [[13]] $ <env: package:stats>
## [[14]] $ <env: package:graphics>
## [[15]] $ <env: package:grDevices>
## [[16]] $ <env: package:utils>
## [[17]] $ <env: package:datasets>
## [[18]] $ <env: package:methods>
## [[19]] $ <env: Autoloads>
## [[20]] $ <env: package:base>rlang package is not part of the search path?To add a package to the search path, use the library()
command.
library(rlang)##
## Attaching package: 'rlang'## The following objects are masked from 'package:purrr':
##
## %@%, as_function, flatten, flatten_chr, flatten_dbl, flatten_int,
## flatten_lgl, flatten_raw, invoke, splicesearch()## [1] ".GlobalEnv" "package:rlang" "package:htmltools"
## [4] "package:knitr" "package:forcats" "package:stringr"
## [7] "package:dplyr" "package:purrr" "package:readr"
## [10] "package:tidyr" "package:tibble" "package:ggplot2"
## [13] "package:tidyverse" "package:stats" "package:graphics"
## [16] "package:grDevices" "package:utils" "package:datasets"
## [19] "package:methods" "Autoloads" "package:base"The tidyverse is a kind of meta-package that loads
several other packages. Note the order in which the new packages are
loaded.
library(tidyverse)
search()## [1] ".GlobalEnv" "package:rlang" "package:htmltools"
## [4] "package:knitr" "package:forcats" "package:stringr"
## [7] "package:dplyr" "package:purrr" "package:readr"
## [10] "package:tidyr" "package:tibble" "package:ggplot2"
## [13] "package:tidyverse" "package:stats" "package:graphics"
## [16] "package:grDevices" "package:utils" "package:datasets"
## [19] "package:methods" "Autoloads" "package:base"tidyverse add to
the search path? Why do you think the developers chose the order they
did?You can find out what environment you are in with
current_env().
current_env()## <environment: R_GlobalEnv>To see what is in an environment, use env_print().
env_print()## <environment: global>
## Parent: <environment: package:rlang>
## Bindings:
## • projects: <tibble[,2]>
## • x: <int>
## • li: <fn>
## • mpg_by_year: <list<tibble[,11]>>
## • mods: <list>
## • starwars: <tibble[,15]>
## • fib_df: <tibble[,3]>
## • fibonacci: <fn>
## • posted: <lgl>
## • .Random.seed: <int>
## • sw2: <tibble[,3]>
## • i: <int>
## • unique_values_safe: <fn>
## • fib: <dbl>
## • unique_values: <fn>
## • talks: <tibble[,2]>Note that while the current environment is usually the global environment, that is is not always the case.
global_env()## <environment: R_GlobalEnv>Let’s write a function that returns the environment that runs during its execution. Note that this is not the global environment.
func_env <- function() {
x <- "what?"
current_env()
}
env_print(func_env())## <environment: 0x7fe574fea558>
## Parent: <environment: global>
## Bindings:
## • x: <chr>Let’s now change the global environment by writing a function called
filter(). This function will just pass its argument to
dplyr::filter() after printing a message to the screen.
filter <- function(.data, ...) {
message(paste("Filtering a", first(class(.data)), "object with", nrow(.data), "rows..."))
dplyr::filter(.data, ...)
}filter() function to find all the human
characters in starwars. Does it work? Why or why not?# SAMPLE SOLUTION
starwars %>%
filter(species == "Human")## Filtering a tbl_df object with 87 rows...Note that our new function filter() is now in the global
environment, which is first in the search path.
env_has(global_env(), "filter")## filter
## TRUEsearch()## [1] ".GlobalEnv" "package:rlang" "package:htmltools"
## [4] "package:knitr" "package:forcats" "package:stringr"
## [7] "package:dplyr" "package:purrr" "package:readr"
## [10] "package:tidyr" "package:tibble" "package:ggplot2"
## [13] "package:tidyverse" "package:stats" "package:graphics"
## [16] "package:grDevices" "package:utils" "package:datasets"
## [19] "package:methods" "Autoloads" "package:base"But there are other environments that contain objects called
filter. We can use map() to search through the
chain of environments.
search_envs() %>%
map_lgl(env_has, "filter")## global package:rlang package:htmltools package:knitr
## TRUE FALSE FALSE FALSE
## package:forcats package:stringr package:dplyr package:purrr
## FALSE FALSE TRUE FALSE
## package:readr package:tidyr package:tibble package:ggplot2
## FALSE FALSE FALSE FALSE
## package:tidyverse package:stats package:graphics package:grDevices
## FALSE TRUE FALSE FALSE
## package:utils package:datasets package:methods Autoloads
## FALSE FALSE FALSE FALSE
## package:base
## FALSEWe can also use find() to show us directly which
environments contain an object that is bound to the name
filter.
find("filter")## [1] ".GlobalEnv" "package:dplyr" "package:stats"filter() function from the
stats package?# SAMPLE SOLUTION
stats::filter()Prompt: If you could clarify one thing about environments, what would it be?