easy_exact_limits

[jonocarroll]

easy_exact_limits
easy_exact_xlim
easy_exact_ylim
teach argument implemented
no tests yet
shortcuts vignette updated

Is this what you had in mind in #1 @AliciaSchep? Happy to add in more functionality if any is needed.

[AliciaSchep]

Not quite what I was thinking, but close. I think the behavior of defaulting to removing the extra space on the lower limit maps well to the motivating use in the issue I created, but is maybe a bit surprising for the function name & current documentation.

I was thinking something more like:

easy_exact_ylim2 <- function(..., teach = FALSE){
  
  lims <- c(...)
  stopifnot(length(lims) == 2)
  
  if (!any(is.na(lims)) && lims[1] > lims[2]) {
    trans <- "reverse"
  }
  else {
    trans <- "identity"
  }
  
  default_expand <- 0.1
  
  expand <- rep(0,4)
  if (is.na(lims[1])){
    expand[1] <- default_expand
  }
  if (is.na(lims[2])){
    expand[2] <- default_expand
  }
  
  # Teach goes here
  
  return(scale_y_continuous(limits = lims, expand = expand, trans = trans))
  
}

Which you would then call

ggplot(mtcars) +
      geom_bar(aes(x = factor(cyl))) +
        easy_exact_ylim2(0,NA)

To get the bar chart at zero effect.

It would basically be just like xlim except for also setting expand to zero for the non-NA value.

To make it easier to just give max or min, I was thinking of making it easy_exact_ylim(..., max = NA, min = NA, teach = FALSE) so you could optionally call easy_exact_ylim(min = 0) but you could also call it similarly to xlim. I think having the functions being a pretty easy drop-in for ylim and xlim would be nice

I also think having the option to alter the expand argument actually takes away some of the usefulness of this function -- part of what is annoying to remember about setting exact limits is that you have to remember what the positions in the expand vector stand for (I literally look this up every time). Having it as an option here would again make you have to potentially figure out that vector, while having the value be zero if the limit is not NA would fit the function name (of making limits exact) and take the burden of dealing with that vector away from the user

[jonocarroll]

That all sounds good. I'll iterate and update.