Compilation error with latest CRAN packages (and R 3.5.2)

Summary:

Compilation error with a test execution of rstan after successful package installation.

Description:

Initially discovered when installing prophet (see https://github.com/facebook/prophet/issues/1027) with a new host. Installations on previous versions of packages was successful, but the latest version fails.

Reproducible Steps:

With R 3.5.2-2 (Enterprise Linux 7 package) on CentOS 7

install.packages('rstan')
library('rstan')
model_code <- 'parameters {real y;} model {y ~ normal(0,1);}'
fit <- stan(model_code=model_code)
mean(extract(fit)$y)

Current Output:

> library('rstan')
Loading required package: ggplot2
Loading required package: StanHeaders
rstan (Version 2.18.2, GitRev: 2e1f913d3ca3)
For execution on a local, multicore CPU with excess RAM we recommend calling
options(mc.cores = parallel::detectCores()).
To avoid recompilation of unchanged Stan programs, we recommend calling
rstan_options(auto_write = TRUE)
> model_code <- 'parameters {real y;} model {y ~ normal(0,1);}'
> fit <- stan(model_code=model_code)
Error in compileCode(f, code, language = language, verbose = verbose) : 
  Compilation ERROR, function(s)/method(s) not created! In file included from /usr/lib64/R/library/BH/include/boost/random/detail/integer_log2.hpp:19:0,
                 from /usr/lib64/R/library/BH/include/boost/random/detail/int_float_pair.hpp:26,
                 from /usr/lib64/R/library/BH/include/boost/random/exponential_distribution.hpp:27,
                 from /usr/lib64/R/library/BH/include/boost/random/gamma_distribution.hpp:25,
                 from /usr/lib64/R/library/StanHeaders/include/stan/math/prim/mat/prob/dirichlet_rng.hpp:5,
                 from /usr/lib64/R/library/StanHeaders/include/stan/math/prim/mat.hpp:292,
                 from /usr/lib64/R/library/StanHeaders/include/stan/math/rev/mat.hpp:12,
                 from /usr/lib64/R/library/StanHeaders/include/stan/math.hpp:4,
                 from /usr/lib64/R/library/StanHeaders/include/src/stan/model/model_header.hpp:4,
                 from file1316ae24e3.cpp:8:
/usr/lib64/R/library/BH/include/boost/
In addition: Warning message:
In system(cmd, intern = !verbose) :
  running command '/usr/lib64/R/bin/R CMD SHLIB file1316ae24e3.cpp 2> file1316ae24e3.cpp.err.txt' had status 1
Error in sink(type = "output") : invalid connection
> 

Expected Output:

Proper calculation of mean value (should be close to 0)

RStan Version:

2.18.2

R Version:

3.5.2

Operating System:

CentOS 7

Extra info

> sessionInfo()
R version 3.5.2 (2018-12-20)
Platform: x86_64-redhat-linux-gnu (64-bit)
Running under: CentOS Linux 7 (Core)

Matrix products: default
BLAS/LAPACK: /usr/lib64/R/lib/libRblas.so

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

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

other attached packages:
[1] rstan_2.18.2          StanHeaders_2.18.1-10 ggplot2_3.2.0        

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.1         magrittr_1.5       munsell_0.5.0      colorspace_1.4-1  
 [5] R6_2.4.0           rlang_0.3.4        tools_3.5.2        parallel_3.5.2    
 [9] pkgbuild_1.0.3     grid_3.5.2         gtable_0.3.0       loo_2.1.0         
[13] cli_1.1.0          withr_2.1.2        matrixStats_0.54.0 lazyeval_0.2.2    
[17] assertthat_0.2.1   tibble_2.1.3       crayon_1.3.4       processx_3.3.1    
[21] gridExtra_2.3      callr_3.2.0        ps_1.3.0           inline_0.3.15     
[25] compiler_3.5.2     pillar_1.4.1       scales_1.0.0       prettyunits_1.0.2 
[29] stats4_3.5.2       pkgconfig_2.0.2   
> 

Reproduce the reported sequence with R 3.5.2 on CentOS 7: install.packages('rstan'), load rstan, and call stan() with the supplied model code. Inspect the compileCode and R CMD SHLIB error output, then verify the fix by confirming that mean(extract(fit)$y) completes and is close to 0.