dplyr source install fails

[jpaxton]

Hi. I'm experiencing the problem described here:

Can't install dplyr in new project

Hi! I tried to install dplyr in a fresh RStudio Cloud project, and failed. As far as I can see, all the dependencies are installed successfully.

Complete output of `install.packages("dplyr")`

> install.packages("dplyr") Installing package into ‘/home/rstudio-user/R/x86_64-pc-linux-gnu-library/3.6’ (as ‘lib’ is unspecified) also installing the dependencies ‘backports’, ‘digest’, ‘zeallot’, ‘utf8’, ‘vctrs’, ‘cli’, ‘crayon’, ‘fansi’, ‘pillar’, ‘purrr’, ‘assertthat’, ‘glue’, ‘magrittr’, ‘pkgconfig’, ‘R6’, ‘Rcpp’, ‘rlang’, ‘tibble’, ‘tidyselect’, ‘BH’, ‘plogr’ trying URL 'http://package-proxy/src/contrib/backports_1.1.4.tar.gz' Content type 'application/x-tar…

Briefly: it looks like R 3.6.0 wants to install dplyr from source, and won't install from the binary package cache. Compilation + installation fails due to lack of memory on the free tier of RStudio Cloud (1GB).

The thread linked above says this issue should be fixed. But I'm still experiencing it, even after deleting and re-creating my project.

Any ideas?

Thanks!
-Joe

[jpaxton]

Full error output follows:

g++: internal compiler error: Killed (program cc1plus)
Please submit a full bug report,
with preprocessed source if appropriate.
See <file:///usr/share/doc/gcc-5/README.Bugs> for instructions.
make: *** [mutate.o] Error 4
/opt/R/3.6.0/lib/R/etc/Makeconf:176: recipe for target 'mutate.o' failed
ERROR: compilation failed for package ‘dplyr’
* removing ‘/home/rstudio-user/R/x86_64-pc-linux-gnu-library/3.6/dplyr’
Warning in install.packages :
  installation of package ‘dplyr’ had non-zero exit status

[jpaxton]

The previous working version compiled for 3.6.0 that was available in mid-May should still be in the package cache, even if a new dplyr version was released recently (which could take a few days to make it into the cache). Right?

[kim4597]

I am facing similar issue today. Anxious to know the solution.

[stevenolen]

Hi!

Because of the holiday, we had a bit of extra sluggishness in ensuring dplyr 0.8.3 hit the package cache. It is now cached, so a new install.packages('dplyr') should result in pulling the cached binary and successful use of dpylr!

[jpaxton]

Yup, all set now. Thanks!