Are there any adverse effect to including library
/require
statements inside of functions that will be called very frequently?
The time used seems rather negligble, but I am calling the function every few minutes and I am wondering if there is any downside to the repetitve require
calls?
note that the function is just a personal util and is not being shared. ie, I am the only one using it
Incidentally, any insight as to why library
is half as slow as require
? I was under the impression they were synonymous.
WithREQUIRE <- function(x) {
require(stringr)
str_detect(x, "hello")
}
WithLIBRARY <- function(x) {
library(stringr)
str_detect(x, "hello")
}
Without <- function(x) {
str_detect(x, "hello")
}
x <- "goodbye"
library(rbenchmark)
benchmark(WithREQUIRE(x), WithLIBRARY(X), Without(x), replications=1e3, order="relative")
# test replications elapsed relative user.self sys.self
# Without(x) 1000 0.592 1.000 0.262 0.006
# WithREQUIRE(x) 1000 0.650 1.098 0.295 0.015
# WithLIBRARY(X) 1000 1.359 2.296 0.572 0.024
Which function is used for loading packages? Explanation: library() function is used to load a package. library() is not useful when we are developing a package since you have to install the package first. A library is a simple directory containing installed packages.
Function packages are basically several external functions and subroutines that are grouped or packaged together.
A dependency is a code that your package needs to run. Dependencies are managed by two files. The DESCRIPTION manages dependencies at the package level; i.e. what packages needs to be installed for your package to work. R has a rich set of ways to describe different types of dependencies.
require
checks whether the package is already loaded (on the search path)
using
loaded <- paste("package", package, sep = ":") %in% search()
and will only proceed with loading it if this is FALSE
library
includes a similar test, but does a bit more stuff
when this is TRUE (including creating a list of available packages.
require
proceeds using a tryCatch
call to library and will create a message .
So a single call to library
or require
when a package is not on the search path may result in library
being faster
system.time(require(ggplot2))
## Loading required package: ggplot2
## user system elapsed
## 0.08 0.00 0.47
detach(package:ggplot2)
system.time(library(ggplot2))
## user system elapsed
## 0.06 0.01 0.08
But, if the package is already loaded, then as you show, require
is faster because it doesn't do much more than check the package is loaded.
The best solution would be to create a small package that imports stringr
(or at least str_extract
from stringr
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With