Suppress false-positive warning for standalone functions

[andrjohns]

Packages exporting standalone functions via rstantools have been receiving the following warning during submission (here and here):

* checking whether package 'rmdcev' can be installed ... WARNING
Found the following significant warnings:
  WARNING: The tools required to build C++ code for R were not found

This warning is a false positive, as the packages compile without issue.

I've identified that this is coming from calling rstan::expose_stan_functions() with rstan 2.21 when not using RStudio (it has also been seen on the forums before)

This PR just adds suppressWarnings(suppressMessages()) to the expose_functions() call

[andrjohns]

@jgabry When you've got the time, would it be much trouble to get this to CRAN? It's blocking the update of the lgpr and rmdcev packages for compatibility with rstan 2.26

[jgabry]

Thanks @andrjohns, I can try to get this to CRAN this coming week.

[andrjohns]

Much appreciated! Also feel free to let me know if there are things I can do to help with CRAN submissions and prep (for this and the other packages as well)

[jgabry]

Much appreciated! Also feel free to let me know if there are things I can do to help with CRAN submissions and prep (for this and the other packages as well)

One thing I've been thinking about recently is that it would be nice to be able to run reverse dependency checks for our various R packages via GitHub actions (maybe using https://github.com/r-lib/revdepcheck#usage?). Probably not to be run every time the regular checks run, but something that can be triggered for major PRs and when prepping a release. What do you think?

[jgabry]

Much appreciated! Also feel free to let me know if there are things I can do to help with CRAN submissions and prep (for this and the other packages as well)

One thing I've been thinking about recently is that it would be nice to be able to run reverse dependency checks for our various R packages via GitHub actions (maybe using https://github.com/r-lib/revdepcheck#usage?). Probably not to be run every time the regular checks run, but something that can be triggered for major PRs and when prepping a release. What do you think?

Actually maybe we can use this: https://github.com/yihui/crandalf

[andrjohns]

Definitely! It's actually something that I had a test run at a couple months back. The main holdup was that it was difficult to parallelise efficiently because each worker would cache the dependencies separately for a given package (as revdepcheck uses the crancache package to locally cache dependencies which can be reused across packages) so things were taking a very long time.

Now that I've worked with the package a bit more for rstan's testing, I think I know a way to get it all working. Another alternative would be to see if Flatiron would be open to us having a reverse-dependency checking workflow on Jenkins. That would be great for testing rstan as part of the downstream tests, and as something to run on demand for the other packages