Problem using nth() with POSIXct date

[steadyfish]

Hi,

Thanks for making this wonderful package dplyr!

I'm facing an issue working with POSIXct dates and it seems very similar to this issue #194. The min() and max() functions work fine with dates, but nth() doesn't seem to retain the class information..

One can replicate this issue with the same example as @manuelreif reported in issue #194, just replacing the function min() with nth().

[romainfrancois]

Capturing modified example:

require(lubridate)
ID     <- rep(letters[1:4],each=5)
date   <-  ymd(paste0(sample(c(1960:2014),length(ID),replace=TRUE),sample(sprintf(fmt="%02d",1:12),length(ID),replace=TRUE),sample(sprintf(fmt="%02d",1:25),length(ID),replace=TRUE)))
number <-  rnorm(length(ID))
# create data.frame
d <- data.frame(ID,date,number)
# use dplyr
d_dplyr <- tbl_df(d)
d_dplyr %>% group_by(ID) %>% summarise(mindate=nth(date, 2))

[steadyfish]

to add, nth() also fails to retain the factor levels when applied on ordered factors.

Example:

 a = rep(seq(1,20,2),3)
 b = as.ordered(a)
 dat = data.frame(a,b)

 is(dat$b)
 [1] "ordered"             "factor"              "integer"             "oldClass"            "numeric"            
 [6] "vector"              "data.frameRowLabels"

 dat1 = dat %>%
   group_by(a) %>%
   summarise(der = nth(b,2))

 is(dat1$der)
 [1] "integer"             "numeric"             "vector"              "data.frameRowLabels"

[hadley]

@romainfrancois is there an implementation of nth() in the hybrid evaluator, or does this just need fixing in the R code?

[romainfrancois]

hybrid definitely has nth. Picking this up now.

[romainfrancois]

Almost there. I only need to deal with factors now.

[nverwer]

It is great that nth handles dates and factors now! Could you make the same changes for first and last, please?

[romainfrancois]

Reopening this to handle first and last as per @nverwer comment.

[nverwer]

Thank you!