compilation failure with openBLAS on Windows #1563
Description
On 2025-05-07, a user reported a compilation failure when switching from R's default BLAS to OpenBLAS on Windows.
@paul-vdb do you happen to have R set up with OpenBLAS? @weizhangstats do you use Windows? If so do you have R set up with OpenBLAS? I'd like to see if we can replicate the error but without any of us having to do anything time-consuming initially.
Here is some detail from the email chain showing NIMBLE's MCMC dll exists but cannot be loaded into R:
Error in inDL(x, as.logical(local), as.logical(now), ...) :
unable to load shared object 'C:/Users/SG14/AppData/Local/Temp/Rtmp2ZxEOB/nimble_generatedCode/P_1_MCMC_05_16_06_35_13.dll':
LoadLibrary failure: The specified module could not be found.
> file.exists("C:/Users/SG14/AppData/Local/Temp/Rtmp2ZxEOB/nimble_generatedCode/P_1_MCMC_05_16_06_35_13.dll")
[1] TRUE
> dyn.load("C:/Users/SG14/AppData/Local/Temp/Rtmp2ZxEOB/nimble_generatedCode/P_1_MCMC_05_16_06_35_13.dll")
Error in inDL(x, as.logical(local), as.logical(now), ...) :
unable to load shared object 'C:/Users/SG14/AppData/Local/Temp/Rtmp2ZxEOB/nimble_generatedCode/P_1_MCMC_05_16_06_35_13.dll':
LoadLibrary failure: The specified module could not be found.
The user's example is here:
library(nimble)
# see ?jags.fit
nfun <- nimbleCode({
for (i in 1:N) {
Y[i] ~ dnorm(mu[i], tau)
mu[i] <- alpha + beta * (x[i] - x.bar)
}
x.bar <- mean(x[])
alpha ~ dnorm(0.0, 1.0E-4)
beta ~ dnorm(0.0, 1.0E-4)
sigma <- 1.0/sqrt(tau)
tau ~ dgamma(1.0E-3, 1.0E-3)
})
## data generation
set.seed(1234)
N <- 100
alpha <- 1
beta <- -1
sigma <- 0.5
x <- runif(N)
linpred <- crossprod(t(model.matrix(~x)), c(alpha, beta))
Y <- rnorm(N, mean = linpred, sd = sigma)
## list of data for the model
ndata <- list(Y = Y, x = x)
## list of constants for the model
nconst <- list('N' = N)
## what to monitor
npara <- c("alpha", "beta", "sigma")
## do mcmc
regmod <- nimbleMCMC(code = nfun, constants = nconst, data = ndata, monitors = npara)
## model summary