renv + pak doesn't work with r-universe repos

[Aariq]

It seems like maybe the repos argument isn't getting passed to pak correctly or something? renv::restore() seems to fail when the lockfile contains R packages installed from r-universe.

To reproduce:

1. Create a new RStudio project
2. install azmetr from r-universe with install.packages("azmetr", repos = c("https://cct-datascience.r-universe.dev", "https://cloud.r-project.org"))
3. Create an R script with library(azmetr) in it
4. Run renv::init() + renv::snapshot()
5. Run options(renv.config.pak.enabled = TRUE)
6. Run renv::restore()

Resulting error message:

Error:                                                                                       
! error in pak subprocess
Caused by error: 
! Could not solve package dependencies:
* azmetr@0.4.0: Can't find package called [email protected].
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
* [email protected]: dependency conflict
Type .Last.error to see the more details.
Traceback (most recent calls last):
6: renv::restore()
5: renv_pak_restore(lockfile = lockfile, packages = packages, exclude = exclude, 
       prompt = prompt, project = project)
4: pak$pkg_install(pkg = remotes, ask = prompt)
3: remote(function(...) get("pkg_install_make_plan", asNamespace("pak"))(...), 
       list(pkg = pkg, lib = lib, upgrade = upgrade, ask = ask, 
           start = start, dependencies = dependencies, loaded = loaded_packages(lib)))
2: err$throw(res$error)
1: base::stop(cond)

sessionInfo()

R version 4.4.1 (2024-06-14)
Platform: x86_64-apple-darwin20
Running under: macOS 15.3.1

Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.4-x86_64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

time zone: America/Phoenix
tzcode source: internal

attached base packages:
[1] stats graphics grDevices datasets utils methods base

loaded via a namespace (and not attached):
[1] compiler_4.4.1 tools_4.4.1 renv_1.1.1 pak_0.8.0

[kevinushey]

Thanks for the bug report -- unfortunately, pak doesn't support installation of specific packages from specific repositories (at least, in a single call). We could probably collect all of the known repositories and provide those when using pak to install / restore packages, but that feels somewhat brittle.

[Aariq]

I see. I guess you'd either have to detect that this was r-universe and do:

pak::pkg_install("azmetr=url::https://cct-datascience.r-universe.dev/src/contrib/azmetr_0.4.0.tar.gz")

perhaps something like this would be a more general solution?

withr::with_options(
  list(repos = c('https://uace-azmet.r-universe.dev', getOption("repos"))),
  pak::pkg_install("azmetr")
)

[kevinushey]

What if azmetr depends on a package that was installed from another repository as well? Or, what if you had multiple repositories providing the same package, but you needed renv to prefer one of those over the other? That's my main worry.

It's probably not a concern in this case, since there's only a single r-universe package, but I think it could be a problem if you had multiple different r-universe repositories in use, or if there was overlap between the packages provided by the repositories.