Support as.Date() conversion within ggplot code

[mkcor]

Sorry, the branch name has become unrelated to the PR.

inherits() checks the class. It's pretty frustrating that I cannot just do

is.null(class(myObject))

!! Class is returned as a string, so it's "NULL", not NULL. Line 52, I'm basically checking for the existence of dimension x.name or y.name. (Don't be confused, the period does not mean anything here!)

The whitespaces are just for styling.

Thanks! @theengineear

[theengineear]

There are two if statements without an else. is something bad going to happen if data.vec isn't set?

[mkcor]

Yes, well, if data.vec is not set, it's NULL and it's still okay because I try() in subsequent operations (https://github.com/ropensci/plotly/blob/master/R/trace_generation.R#L56, etc.). So the exceptions are handled.

My initial intention was to leave line 50

data.vec <- l$data[[col.name]]

and follow with

if (is.null(data.vec)) {
  if (!inherits(g$data[[paste0(a, ".name")]], "NULL") && (class(g$data[[a]]) != class(g$data[[paste0(a, ".name")]]))) {
    g$data[[a]] <- g$data[[paste0(a, ".name")]]
    data.vec <- g$data[[a]]
  }
}

The else being, well, leave it alone (it will remain NULL). Do you think this would convey more meaning?

The two ifs really mean one condition (i.e., the two columns are of different non-null classes) but I split them for readability, also to save one check if the first condition is not even met.

[theengineear]

@mkcor , if the logic is ok with it being left undefined, its seems good : D

🎉