Error in model fit: C++14 standard requested but CXX14 is not defined

Summary:

Error message relating to C++compiler when running a linear model fit.

Description:

I am new to Stan and RStan and trying out the tutorial by Max Farrell and Isla Myers-Smith on:
https://ourcodingclub.github.io/2018/04/17/stan-intro.html

The step by step process is shown below under 'Reproducible Steps'.

When I run the stanc() it compiles fine. But when I fit the linear model I receive the following error message:
Error in compileCode(f, code, language = language, verbose = verbose) :
Compilation ERROR, function(s)/method(s) not created! Error in .shlib_internal(commandArgs(TRUE)) :
C++14 standard requested but CXX14 is not defined
Calls: -> .shlib_internal
Execution halted
In addition: Warning message:
In system(cmd, intern = !verbose) :
running command 'C:/PROGRA1/R/R-351.1/bin/x64/R CMD SHLIB file55c71d73e8e.cpp 2> file55c71d73e8e.cpp.err.txt' had status 1

I've searched everywhere for possible solutions. I reviewed the following issue posted by yanys7
https://github.com/stan-dev/rstan/issues/565

As suggested by bgoodri I created a ~/.R/Makevars file and inserted the following:
CXX14 = g++

However the problem still persists.

I don't know if this is a bug or if I'm doing something wrong.
Thanks in advance for any help you can provide.

Reproducible Steps:

This follows almost exactly the steps shown in the tutorial.

Sys.setenv(USE_CXX14 = 1)
library(rstan)
library(gdata)
library(bayesplot)
rstan_options(auto_write = TRUE)
options(mc.cores = parallel::detectCores())
seaice <- read.csv("seaice.csv", stringsAsFactors = F)
colnames(seaice)<-c("year", "extent_north", "extent_south")
x <- I(seaice$year - 1978)
y <- seaice$extent_north
N <- length(seaice$year)
stan_data <- list(N = N, x = x, y = y)
write("// Stan model for simple linear regression
data {
      int <lower = 1> N;
      vector[N] x; 
      vector[N] y;
      }
      
      parameters {
      real alpha; 
      real beta; 
      real<lower = 0> sigma;
      }
      
      model {
      y ~ normal(alpha + x * beta, sigma);
      }
      generated quantities {
      }, 
"stan_model1.stan")
stanc("stan_model1.stan")
stan_model1 <- "stan_model1.stan"
fit <- stan(file = stan_model1, data = stan_data, warmup = 500, iter = 1000, chains = 4, thin = 1)

Current Output:

The output from stanc("stan_model1.stan") is below:
$status
[1] TRUE

$model_cppname
[1] "model55c23451a60_stan_model1"

$cppcode
[1] "// Code generated by Stan version 2.17.0\n\n#include <stan/model/model_header.hpp>\n\nnamespace model55c23451a60_stan_model1_namespace {\n\nusing std::istream;\nusing std::string;\nusing std::stringstream;\nusing std::vector;\nusing stan::io::dump;\nusing stan::math::lgamma;\nusing stan::model::prob_grad;\nusing namespace stan::math;\n\ntypedef Eigen::Matrix<double,Eigen::Dynamic,1> vector_d;\ntypedef Eigen::Matrix<double,1,Eigen::Dynamic> row_vector_d;\ntypedef Eigen::Matrix<double,Eigen::Dynamic,Eigen::Dynamic> matrix_d;\n\nstatic int current_statement_begin__;\n\nstan::io::program_reader prog_reader__() {\n stan::io::program_reader reader;\n reader.add_event(0, 0, "start", "model55c23451a60_stan_model1");\n reader.add_event(19, 19, "end", "model55c23451a60_stan_model1");\n return reader;\n}\n\nclass model55c23451a60_stan_model1 : public prob_grad {\nprivate:\n int N;\n vector_d x;\n vector_d y;\npublic:\n model55c23451a60_stan_model1(stan::io::va...

$model_name
[1] "stan_model1"

$model_code
[1] "// Stan model for simple linear regression\ndata {\n int <lower = 1> N; // Sample size\n vector[N] x; //Predictor\n vector[N] y; // Outcome\n}\n\nparameters {\n real alpha; //Intercept\n real beta; // Slope\n real<lower = 0> sigma; //Error SD\n}\n\nmodel {\n y ~ normal(alpha + x * beta, sigma);\n}\n\ngenerated quantities {\n} // The posterior predictive distribution"
attr(,"model_name2")
[1] "stan_model1"

Expected Output:

If applicable, the output you expected from RStan.

RStan Version:

RStan version 2.17.3

R Version:

R Version 3.5.1

Operating System:

Windows 8 64bit

Start by reproducing the supplied R script with RStan 2.17.3 on Windows 8, then inspect the ~/.R/Makevars CXX14 setting and the compiler invocation shown in the error. The issue is resolved when the linear-model fit compiles and runs without the “CXX14 is not defined” error.