pkgcache and Posit Package Manager on Linux

PPM and P3M

Posit Package Manager (PPM) is a Posit product to help organizations organize R and Python packages.

Posit Public Package Manager (P3M) is a free instance of PPM.

pkgcache deals with all PPM instances the same way currently, including the P3M instance.

In the following, by PPM we mean any PPM instance.

Why PPM?

Fast installation. PPM has binary packages on Linux, for many distributions, for the last 5 R releases.
New packages on older R versions. On Windows, new package versions are available as binary packages, on older R versions as well.
Time traveling. Use an older state of the repository, as it was on a specific date.

pkgcache and PPM on Linux

For a better user experience pkgcache handles PPM repositories specially on Linux.

Call pkgcache::ppm_has_binaries() to decide if PPM builds binaries for your platform:

pkgcache::ppm_has_binaries()

## [1] TRUE

You can use pkgcache::current_r_platform() to see if pkgcache detects your platform correctly:

pkgcache::current_r_platform()

## [1] "x86_64-pc-linux-gnu-ubuntu-22.04"

Setting up PPM

Set the repos option to a PPM URL, as suggested by the PPM setup web page. E.g. see https://packagemanager.posit.co/client/#/repos/2/overview for P3M.

You can also set the PKGCACHE_PPM_URL environment variable to the base URL of a PPM instance (e.g. https://packagemanager.posit.co for P3M), and the PKGCACHE_PPM_REPO environment variable to the default PPM repository. The pkgcache::repo_resolve() and pkgcache::repo_add() functions use these environment variables (if they are set) to construct PPM URLs. See ?ppm_repo_url for details.

PPM repository URLs

A PPM repository URL for Linux binary packages has the following form:

{base}/{repo}/__linux__/{binary_url}/{id}

where {base} is the base URL of the PPM instance, {repo} is a PPM repository, {binary_url} is the code name of a release of a Linux distribution (e.g. rhel9), and {id} is the identifier (or date) of a PPM snapshot or latest for the latest snapshot.

ppm_platforms() lists platforms supported by a PPM instance:

pkgcache::ppm_platforms()

## # A data frame: 27 × 6
##    name        os    binary_url  distribution release binaries
##    <chr>       <chr> <chr>       <chr>        <chr>   <lgl>   
##  1 centos7     linux centos7     centos       7       TRUE    
##  2 centos8     linux centos8     centos       8       TRUE    
##  3 rhel9       linux rhel9       rockylinux   9       TRUE    
##  4 opensuse15  linux opensuse15  opensuse     15      TRUE    
##  5 opensuse152 linux opensuse152 opensuse     15.2    TRUE    
##  6 opensuse153 linux opensuse153 opensuse     15.3    TRUE    
##  7 opensuse154 linux opensuse154 opensuse     15.4    TRUE    
##  8 opensuse155 linux opensuse155 opensuse     15.5    TRUE    
##  9 opensuse42  linux opensuse42  opensuse     42.3    TRUE    
## 10 rhel7       linux centos7     redhat       7       TRUE    
## # ℹ 17 more rows

The binary_url column contains the code name that you need to use in the repository URL.

ppm_snapshots() lists all PPM snapshot dates and ids:

pkgcache::ppm_snapshots()

## # A data frame: 1,000 × 2
##    date                   id
##    <dttm>              <int>
##  1 2019-02-28 00:00:00   148
##  2 2019-03-04 00:00:00   149
##  3 2019-03-07 00:00:00   150
##  4 2019-03-11 00:00:00   151
##  5 2019-03-14 00:00:00   152
##  6 2019-03-18 00:00:00   153
##  7 2019-03-21 00:00:00   154
##  8 2019-03-25 00:00:00   155
##  9 2019-03-28 00:00:00   156
## 10 2019-04-01 00:00:00   157
## # ℹ 990 more rows

You can use the snapshot ids or dates in the PPM repository URL to select a particular snapshot. E.g. these two repository URLs are equivalent:

https://packagemanager.posit.co/cran/__linux__/centos7/5
https://packagemanager.posit.co/cran/__linux__/centos7/2017-10-10

Linux platforms

pkgcache generalizes R’s “machine-vendor-OS” platform triplet on Linux to include the distribution and release as well. E.g. the platform of the machine building this document is:

pkgcache::current_r_platform()

## [1] "x86_64-pc-linux-gnu-ubuntu-22.04"

PPM Linux binary packages

R does not have built-in support for binary Linux repositories, so PPM offers binary Linux packages as source packages. This is possible because R recognizes binary packages at installation time, and handles them appropriately.

pkgcache detects binary PPM repositories, and constructs their platform string:

print(
  pkgcache::meta_cache_list(package = c("ggplot2", "pkgcache"))[, c("package", "mirror", "platform")],
  width = Inf
)

## # A data frame: 4 × 3
##   package  mirror                                                     
## * <chr>    <chr>                                                      
## 1 ggplot2  https://packagemanager.posit.co/cran/__linux__/jammy/latest
## 2 pkgcache https://packagemanager.posit.co/cran/__linux__/jammy/latest
## 3 ggplot2  https://cran.rstudio.com                                   
## 4 pkgcache https://cran.rstudio.com                                   
##   platform                        
## * <chr>                           
## 1 x86_64-pc-linux-gnu-ubuntu-22.04
## 2 x86_64-pc-linux-gnu-ubuntu-22.04
## 3 source                          
## 4 source

Metadata discrepancies

Occasionally, the PPM metadata can be outdated, and pkgcache will wrongly think that PPM has a binary for a package, whereas it does not, or the opposite might also be true.

When pkgcache downloads a package as an update to the package cache, it detects from the PPM response headers whether the downloaded package is indeed of the expected type. If not, then pkgcache updates its knowledge about the package.

Supported R versions

Different PPM instances may support different R versions. The public P3M instance supports the last five R releases. pkgcache will automatically check if your configured PPM instance supports your R version, and ppm_has_binaries(), repo_set() will act accordingly. You can use the pkgcache::ppm_r_versions() function to list the R versions supported by your instance:

pkgcache::ppm_r_versions()

## # A data frame: 5 × 1
##   r_version
##   <chr>    
## 1 4.3      
## 2 4.2      
## 3 4.1      
## 4 4.0      
## 5 3.6

Note that these version numbers do not include the patch version. E.g. 4.2 means that all 4.2.x R versions are supported.

pkgcache and PPM on Windows

A PPM instance can have support for Windows binaries, the public P3M instance does support them.

The R package system supports Windows binaries out of the box, so setting up PPM on Windows simply means setting the repos option to a PPM repository. E.g. for the public P3M this would be something like:

options(repos = c(
  RSPM = "https://packagemanager.posit.co/cran/latest",
  CRAN = getOption("repos")[["CRAN"]]
))

Instead of latest you can use a different snapshot, of course, and you can also use repo_resolve() to construct the URL for your preferred snapshot.

Gábor Csárdi

2024-04-04