同一R库中能否使用多版本R包？如何指定版本加载？

Absolutely, you can access multiple versions of the same R package when running code under a shared production user account. Below are practical tools and workarounds to achieve the exact library(my_package, version='1.0.1') behavior you're looking for:

1. Use the versions Package (Direct Version-Specific Loading) #

The versions package isn't just for installing specific package versions—it also supports loading a targeted version directly, as long as you've installed the different versions into separate library locations first.

First, install the package versions you need into distinct library paths:

# Install my_package 1.0.1 to a dedicated library
install.packages("my_package", version = "1.0.1", lib = "/prod/libs/my_package_v1.0.1")
# Install my_package 1.1.0 to another library
install.packages("my_package", version = "1.1.0", lib = "/prod/libs/my_package_v1.1.0")

Then use the versions::library() function to load the exact version you want:

versions::library(my_package, version = "1.0.1", lib.loc = "/prod/libs/my_package_v1.0.1")

If you want to simplify this to match your desired syntax, wrap it in a custom function:

library_version <- function(pkg, version) {
  lib_path <- file.path("/prod/libs", paste0(pkg, "_v", version))
  versions::library(pkg, version = version, lib.loc = lib_path)
}

# Now use it like this
library_version("my_package", "1.0.1")

2. Manual Library Path Management with Base R #

Base R's library() function has a lib.loc parameter that lets you specify exactly which library directory to load a package from. This is a lightweight approach if you prefer not to add extra dependencies.

First, organize your package versions into versioned library directories:

# Create dedicated library folders (run once)
dir.create("/prod/libs/v1.0.1", recursive = TRUE)
dir.create("/prod/libs/v1.1.0", recursive = TRUE)

# Install each version to its folder
install.packages("my_package", version = "1.0.1", lib = "/prod/libs/v1.0.1")
install.packages("my_package", version = "1.1.0", lib = "/prod/libs/v1.1.0")

Then load the specific version by pointing to its library:

# Load version 1.0.1
library(my_package, lib.loc = "/prod/libs/v1.0.1")

# Verify the version
packageVersion("my_package")

3. Isolate Environments with renv #

For more robust environment isolation (great if different code scripts need entire sets of package versions), renv lets you create separate project environments under the same production user. Each environment can have its own version of my_package.

1. Create a dedicated environment for each version:

# Set up an environment for my_package 1.0.1
renv::init(project = "/prod/envs/env_v1.0.1")
renv::install("my_package@1.0.1")
renv::snapshot()

# Set up another for 1.1.0
renv::init(project = "/prod/envs/env_v1.1.0")
renv::install("my_package@1.1.0")
renv::snapshot()

2. When running code, activate the desired environment first:

# Activate the 1.0.1 environment
renv::activate("/prod/envs/env_v1.0.1")
library(my_package) # Loads 1.0.1 automatically

# When switching versions, activate the other environment
renv::activate("/prod/envs/env_v1.1.0")
library(my_package) # Now loads 1.1.0

Key Notes #

• Ensure the production user has read/write access to all library directories or renv environments you create.
• Tools like miniCRAN or cranium can be used alongside these methods to manage package repositories and ensure you install the exact versions you need.

内容的提问来源于stack exchange，提问作者statquant