Open athowes opened 3 weeks ago
I'm dodging the majority of your good questions here to think a little more, but to answer a few specific things:
I think it would be helpful to see a document put together of the expected workflow.
You've (rightly) asked for this a few times and it's an important thing we need to provide. Bear with me for a few more PRs until we have an actual pipeline-running function and let's revisit. I'm trying to avoid documenting ahead of the code.
I think it's important for documentation and development of the package that one could use it in a more basic environment.
Can you expand on this a bit? We have in the readme:
But it sounds like you have a broader scope in mind. What use-cases are you thinking about? In particular, when would one want to use our package over vanilla EpiNow2?
In PR #26 we added modelling functionality, and chose not to add defaults for arguments like the prior:
_Originally posted by @zsusswein in https://github.com/CDCgov/cfa-epinow2-pipeline/pull/26#discussion_r1755154014_
I think this is a reasonable choice / likely correct.
I do think there could be a discussion to be had about the extent to which we are making this package "config-driven" versus "interactive". One could frame "interactive" as in "working like ordinary R package functions". And then one could argue that the pipeline should do the work of being config driven, and that functions in the R package could be functional from an "interactive" perspective / ordinary R package perspective. I do think there are instances where we / our collaborators might interact with the package in an interactive mode.
Sorry, this is quite a non actionable issue! I'd be interested in others broader perspectives on this as a broader design question. As I say, I think the choice to make things quite "config-driven" is reasonable.
I think it would be helpful to see a document put together of the expected workflow. Perhaps another part of this "package specificity" question is the extent to which this package can be used outside of an Azure / CFA-specific environment. I think it's important for documentation and development of the package that one could use it in a more basic environment.