Failed to build Lotka Vollterra on Mac OS / clang about example-models

[bob-carpenter]

Thanks much for following up. Sounds like the RStudio team is making this a high priority for a fix after their 1.3 release.

Do you know if this is also a problem with the terminal version of R 4.0, too?

Heads up to the R interface devs: @bgoodri, @jgabry, and @rok-cesnovar, as this may impact RStan and CmdStanR.

[bob-carpenter]

I think it's an install problem with Stan on Mac OS X. We've had a ton of problems with Catalina specifically. I have them ironed out in my Mac OS X Catalina install, and everythig builds OK for me.

[bob-carpenter]

Sorry, meant to go on. Can you try to fit a simple Stan model in the same environment? I'm afraid I'm not sure where the up-to-date how-to-install-with-Catalina instructions are.

[mitzimorris]

what interface are you using? RStan?

[marcmenem]

Thanks Bob. I see it is definitely related to:
https://github.com/stan-dev/rstan/wiki/Catalina-problems

It is confusing because simple models work on a fresh R Kernel, but eventually they can start failing (which is wy I though the issue was with some specific functions.

I'll try to get this sorted out on my install, but any pointers are welcome.

[marcmenem]

Yes, using rstan

[marcmenem]

FWIW, upgrading to R 4.0 seems to fix the exceptions issue and I was able to run the example, but there I had to turn off parrellel computing because it failed to connect.

[ssp3nc3r]

FWIW, upgrading to R 4.0 seems to fix the exceptions issue and I was able to run the example, but there I had to turn off parrellel computing because it failed to connect.

It's an issue with RStudio and R4.0:

rstudio/rstudio#6692

You should be able to run parallel chains from the command line, and maybe from RStudio jobs( haven't confirmed the latter yet).

[ssp3nc3r]

Do you know if this is also a problem with the terminal version of R 4.0, too?

No, terminal/command line version is fine, as is R Gui. And @bgoodri mentioned that jobs within RStudio may also be fine.