Consider how a FIT R universe page can supplement the FIT website.

[k-doering-NOAA]

Some information about how other communities are using R universe

[eeholmes]

We use it for our time-series packages: https://atsa-es.r-universe.dev/packages. So easy to set up and makes it much much easier for people to install your GitHub packages esp if they include C++ code (e.g. TMB) because they don't have to build them. Works like installing from CRAN.

[k-doering-NOAA]

We do have this experimental FIT one.

I think the other advantages is you need the whole r-tools chain for github pkgs with c++ code if you install from r universe.

[k-doering-NOAA]

I attended a R open sci coworking session with Jereon Ooms, the creator and developer of the R universe. Here's some things I learned:

• R universe does not archive binaries, so only the currently built binary for the package is available. However, the snapshot API (e.g., available on this page) can be used to download a zip file with your built binaries (and other information), that you can distribute (e.g., on GitHub releases). More info in this FAQ about package archiving
• There are branch and subdir fields that can be included in the package.json file. I'm struggling to find documentation, but the subdir allows you to specify a subdirectory that the R package is in, and the branch field allows you to specify a branch or tag that the binary should be built from. Pattern matching also works: for example, you could use *release I think builds any release created on GitHub?
• You can see the activity in a repository over the past year on a packages' individual page. It will also show when releases are created.
• There is a data set api (example), which seems like useful way for people to get datasets from R packages
• Monitoring the builds page could be a way for us or for package maintainers to make sure that their package builds aren't failing.

[k-doering-NOAA]

Recent value added: devtools::install_github() isn't working on NOAA laptops, and pak::pak() required the package build tool chain. Installation from R universe does not, so the user was able to install from R universe as a "quick fix" way to get the package nmfspalette, which is not on CRAN.

R universe can be a helpful tool for users and developers of the toolbox; We could provide instructions on how to install using R universe on the R package pages?

[k-doering-NOAA]

I think there are some clear uses of R universe that could be helpful. I think when developing the FY25 plan, we can consider specific action items based on what we know about R universe. For example:

• Committing to maintain an R universe for the FIT and advertise it(we have one, but it is not well advertised)
• Add install instructions using R universe to all of the FIT R packages