Can't install RcppEigen or BayesFactor in Posit Cloud

[Originally posted as part of this thread, but was advised to post specifically to Posit Cloud instead: Cannot install RcppEigen in R]

I (and my students) cannot install RcppEigen or BayesFactor on Posit Cloud, though both install and work just fine on my RStudio desktop. I have already (repeatedly):

  • Restarted the R session
  • Deleted the lock file mentioned in the original post
  • Restarted my browser (and laptop)
  • Tried installing from archive files (.tar.gz)

The RcppEigen install hangs up every time at the same spot in a long list of messages and warnings:

./inst/include/Eigen/src/Core/DenseCoeffsBase.h:56:30: warning: ignoring attributes on template argument ‘Eigen::internal::packet_traits<double>::type’ {aka ‘__m128d’} [-Wignored-attributes]

After several minutes of being stuck there, I am kicked out with a popup message about "Login expired or signed out from another window," even when the active window is literally the only application of any kind open on my laptop.

I get the same thing with BayesFactor, likely because it depends on RcppEigen.

Any suggestions would be much appreciated - thanks!

Based on your thread you linked it looks like you are R 4.4, have you tried R 4.5.2? I was able to successfully install on that version of R.

Hi @Alyssa_Gibbons, sorry you've had so much trouble with this. I was able to reproduce an issue like you describe using the default resource settings (1 GB of memory), but was able to install RcppEigen in Posit Cloud successfully if I first increased the resource allocation on the project to 6 GB. Somewhere between 1 and 6 would likely work too.

I am on 4.5.2, though I tried cycling through a few previous versions along the way too! Still no luck.

Thanks - I'll give that a try, but unfortunately it won't solve the problem for my students who are using free accounts. It's good to have a likely explanation, though.

Have you tried using binary packages from P3M, i.e. https://packagemanager.posit.co? Then you don't need to compile packages at all. (Well unless some packages are not available at P3M.)

I suspect that if you create a project and install the required packages, and either place it in a shared space with your students where they can copy it or make it an assignment, then your students will be able to open new projects starting from where you left off, with these packages ready to go.

We had an issue that caused some packages to be installed from source when they should not have been. If you try to install RcppEigen now, it should install the binary package and go much faster. If you don't see that, can you let me know.

Thanks for bringing this up!
sam

Thanks so much, everything is installing successfully now!

Appreciate the tip! Most of my students use desktop RStudio right now, so this is only affecting a few, but that's worth exploring for the future.

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.