Avoid global variables

[ateucher]

Refactor dplyr and ggplot code using standard evaluation "escape hatches"

[steffilazerte]

So far I've addressed this by:

• using aes_string() in ggplot2 function calls
• using .data$variable notation in tidyverse function calls (i.e. mutate, group_by, etc)

Using .data$variable avoids the possibility of accidentally using a variable in the global environment but it still results in one undeclared global variable: .data

The only other way (that I've found) to avoid global variables entirely is to use the rlang package, for example:

g %>% 
  dplyr::group_by(rlang::UQS(rlang::parse_expr("Well_Num")))

This seems overly cumbersome and requires another package. It's entirely possible there's an easier way, but I haven't been able to figure it out, as of yet.

I would suggest either:

• Using the globalVariables() function to declare all NSE variables we use (removes Notes from the output of check)
• Use .data and aes_string and be satisfied with one undeclared global variable (.data) [Currently implemented]

[ateucher]

That's great! Yeah, I think the rlang way is the new way to do things, but I think since these in general are not user-supplied variables (they are column names that must be in the data, correct?) the .data approach seems entirely reasonable. I also noticed here that you can use @importFrom rlang .data to avoid .data being identified as a global variable. (And dplyr already imports rlang, so I'm not too worried about adding it to our Imports:.

[steffilazerte]

Fixed: 8787364