Daniel Turek

@perrydv Naming the argument `required` sounds good. My personal feeling is to make the default be `required = FALSE`, but I do understand about changing the API. I'll defer to...

@perrydv Since this PR seems to be merged into `ADoak` and `AD-rc0` branches, can we safely close this PR?

Modifying `minMaxHandler` to error out (with an informative message) in the presence of any named arguments sounds like a sound first step. The error message could even mention "in particular,...

Error trap in what manner? One logical perspective is that `pi` should be a valid variable name in models (for stochastic, or RHS nodes) as in this example, without throwing...

Maybe, perhaps, issuing a `Note`, at the time of model building: ``` [Note]: The use of "pi" as a variable name is detected in model. This will represent a model...

@DRJP Only noting that your minimally reproducible example (using `rinvgamma`) runs fine for me (no errors, and output appears correct), using nimble version 0.6-8 (installed from CRAN), running on OSX....

@DRJP Nice detective work, thanks for following-up.

PR #1299 saves the compiled model object, and uses it for calculations in the case of predictive loss. It's still an open discussion regarding whether predictive loss should be given...

@paciorek Noted, that sounds good. I could fix the exponentiation computational approach at some point, as well. For now, I'll leave this open. I'm going to merge #1299, which might...

@paciorek I think it was the case that some historical-types of model structures, backwards indexing can just "arise" somewhat naturally, and in those cases, it should be ignored... that was...