Issue of lazy evaluation with dev magrittr

[lionel-]

Hello, I'm seeing an issue with the development version of magrittr which switches to a lazy evaluation approach. It causes this error in your package:

> aframe <- data.frame(zed = runif(100))
> set_to_zero <- . %>% mutate(zed = 0)
> aframe %>% scion(zed >0.5, false_fun=set_to_zero) %>% mutate(zed=1) %>% graft
Error in .pop() : 
  No more scions on stack. Perhaps you meant to specify the "data2" argument to graft()?
Calls: %>% -> graft -> .pop

That's because the pipeline is now evaluated as if it were nested, instead of linearly:

graft(mutate(scion(aframe, zed > 0.5, false_fun = set_to_zero), zed = 1))

Because of lazy evaluation, the input to graft has not been evaluated yet when you check if data2 is missing. One way to fix this is to force() the input like this:

--- a/R/tabe.R
+++ b/R/tabe.R
@@ -62,6 +62,7 @@ scion <- function(.data, ..., false_fun, false_name, false_env){
 #' @importFrom dplyr intersect
 #' @importFrom dplyr full_join
 graft <- function(.data, combine_fun, data2){
+  force(.data)
   if(missing(data2))
     data2 <- .pop()

We're planning to release magrittr in one month. It'd be great if you could update your package :)

Note that the future native pipe in the next version of R will use this same approach, so by fixing your package for the new magrittr behaviour you'll make it usable with the native pipe as well.

[restonslacker]

Thank you for the report and the suggested fix. I will work on this week. it's been quite a while since i've worked on this package so i suspect it will need some other updates to remain in compliance with CRAN.

[restonslacker]

package submitted to CRAN. awaiting response.

[restonslacker]

accepted on cran

[lionel-]

Thanks!