Address warnings and notes (as much as possible) from `R CMD check` about mmrm HOT 6 CLOSED

@dinakar29 thanks for this, unfortunately I cannot reproduce the warning you get for the compiled code on my local machine. Is it possible for me to reproduce this myself on enableR? What are the steps for that? (Only then I can try to change something to fix it)

from mmrm.

@dgkf fyi - we have been diligent here trying to fix check results as much as possible, but on our internal R 4.0.4 server there might still be a message, which originates outside of the mmrm package code it seems, and is not shown during checks on R 4.2.

from mmrm.

Thanks @danielinteractive - I've seen this in other compiled packages and it seems to be raised in error. If you do see anything problematic about this warning that would be reason to be concerned, we'd appreciate your input - you have more background with packaging C code than I do.

I think this is a benign warning just due to the system (C compiler differences perhaps?). Even if it is problematic, as far as I can tell, the worst possible outcome would be that the process is terminated, which might be frustrating for users, but is rather unlikely to produce any erroneous results. That said, we haven't seen it actually abort in our exploration so far.

from mmrm.

Thanks a lot @dgkf-roche , that is good to hear. I also think it is a false positive warning in this case and I guess it could even be a bug in R CMD check itself that has been fixed in more recent R versions.

I am further working now on fixing the relevant warning about the registration of Call routines.

from mmrm.

If there's a way to register the routines that doesn't require acrobatics around TMB, that's awesome. If not, then it's a tradeoff between trying to appease R CMD check and wanting to use more widely accepted tools as they were intended. This is similar to building models with rstan, where the generated code produces a few known R CMD check notes.

In either case, I think there's value in staying consistent with other TMB tools, even if that comes with some R CMD check flags. I don't see a problem with that if there's good justification for it.

from mmrm.

Thanks @dgkf - so I took the code now more or less from glmmTMB in the pull request here and it looks relatively concise, and what is important depends on TMB to define the names for the Call routines to register. It would be awesome if you could review that. (No problem if you don't have time, just let me know)

from mmrm.