Mitzi Morris

I couldn't agree more. I looked at the classes in `src/cmdstan/argument` - the code is a bit opaque. are these object mutable? not sure how to change the `algorithm` argument...

OK, will look into messing with `valid_arguments`. since we've got a fix for this in CmdStanPy, not a huge priority. just another example of why we really need to change...

> it seems confusing to change what users requested to something else because what they requested is incoherent. the user's request was vague, but then defaults were applied, resulting in...

thanks for the clear and detailed report. sounds like a bug. will investigate.

standalone GQ currently only checks that the total number of parameter in the model matches the total number of parameters in the supplied fitted_params file. not sure what it would...

note this: https://github.com/stan-dev/stan/issues/2534

PR #992 improved error messaging by sending error messages to stderr instead of stdout. however, the problem with return codes remains; leaving issue open for further improvements.

suggestion would be to follow this: https://shapeshed.com/unix-exit-codes/ > What exit code should I use? > > The Linux Documentation Project has a list of reserved codes that also offers advice...

given the above, all stan::services return codes should map to either 1 or 126. recommend 1.

agreed - warnings might be too strong for iterations at max treedepth. the problem is this: we don't have enough information in the output to distinguish between whether or not...