Closed andrjohns closed 1 year ago
Thanks so much @andrjohns ! You read my mind: last week I fixed these typos in a local branch, but haven't committed them because I wasn't sure what to do about Makevars. So it's great to know that the makevars files should no longer be included.
However, I get a file too big
compilation error on Windows (on Ubuntu I think it's alright) with R 4.2.2, rstan and StanHeaders 2.26.13.
When digging into this last week, I saw that someone on Stan Discourse had a similar problem. There was also a post by Frank Harrel on this. Is there some guidance on what the Makevars should be for developers using rstantools? (Especially Makevars.win) I assume I should have something in ~/.R/
? Maybe I missed some other necessary step in updating the package to new R/rstan versions?
Any suggestions would be great. I've been trying to figure this out on my own but it's a bit confusing because the instructions concerning Makevars have changed over years, so sometimes Google just turns up some suggestions that no longer apply.
Sorry for missing this! I've updated the PR with the fix for this error on Windows
Would you be able to merge this and release to CRAN when you can? Thanks!
Hi @andrjohns it seems like this configure.win
fix nearly does what's needed, but there is one last stumble.
If I run pkgbuild::compile_dll()
on my Windows machine and look at Makevars, I can see PKG_CXXFLAGS += -Wa,-mbig-obj
added to Makevars.win
. But a few seconds later, presumably when config file runs rstantools::rstan_config()
in the last line of your fix, the new flags disappear. I use rstantools 2.2.0. Am I doing this wrong?
Sorry about that! I've updated the configure.win
file and verified that it works locally for me under Windows and 2.26, can you check on your system?
Thanks, this all works very nicely now. I will merge this but open a separate issue to ask you a couple questions, basically I'd be really grateful if you could give me some more info on this problem.
PS: FYI, this will go to CRAN in a few days, there is another critical bug that CRAN flagged recently so I need to look into that.
This PR updates some minor typos in your Stan code (using the
_lpmf
suffix in a sampling statement), and removes files that are automatically provided by rstantools