I have just submitted a revised version of survHE on CRAN — it should be up very shortly. This will be version 1.0.64 and its main feature is a major restructuring in the way the rstan/HMC stuff works.
Basically, this is due to a change in the default C++ compiler. I don’t think much will change in terms of how survHE works when running full Bayesian models using HMC, but R now compiles it without problems. After the advice of Ben Goodrich, I have also modified the package so that it compiles the Stan programs serially as opposed to gluing them all together, which should optmise the use of the memory.