Spooky errors on CRAN winbuilder for rstan dependent package

Is anyone else getting strange problems on winbuilder with rstan model compiles? All checks on other windows machines I can find are ok, but on winbuilder I get errors everywhere as if it is using either an older version of the model, data, or just can’t find the data objects when called by rstan. Like this:

Error in new_CppObject_xp(fields$.module, fields$.pointer, ...) : 
  Exception: variable does not exist; processing stage=data initialization; variable name=tipredsdata; base type=matrix_d  (in 'model_ctsm' at line 308)

Tag @hsbadr

@Charles_Driver Which version of RStan / stanc3? In general, this happens when one or more data objects aren’t parsed correctly from the C++ code. For instance, stanc3 has added _data suffix to internal names, which has been addressed in the experimental branch of RStan

Can you share a reproducible example? I’d also test it using:

remove.packages(c("StanHeaders", "rstan"))
remotes::install_github("stan-dev/rstan/StanHeaders@experimental", upgrade = "always", force = TRUE)
remotes::install_github("stan-dev/rstan/rstan/rstan@experimental", upgrade = "always", force = TRUE)

Well that’s the fun part, only the owners of win-builder upload page (i.e CRAN) can reproduce it, so I was wondering if other package devs experience similar. No point playing with github stuff, CRAN only.

1 Like

turned out to be some kind of makevars issue I don’t fully understand but resolved by reverting to a good copy, for any far future readers.