CODEWITHSUNDEEP
rif-statementdplyrnulldata.table
Nakx
Nakx

Reputation: 1600

ifelse(length(date)==1, date, otherdate) differs between base R, dplyr and data.table

I am trying to test if elements of a date vector are of length 0.

this works fine using base R, but returns POSIXct as numeric:

a<-numeric(0)

>ifelse(length(a)==1,a,as.POSIXct("2017-12-26 17:53:53 AEDT"))
[1] 1514271233

a solution to keep the class is to use dplyr or data.table, but they both return errors:

>dplyr::if_else(length(a)==1,a,as.POSIXct("2017-12-26 17:53:53 AEDT"))
Error: `true` must be length 1 (length of `condition`), not 0
Run `rlang::last_error()` to see where the error occurred.

>data.table::fifelse(length(a)==1,a,as.POSIXct("2017-12-26 17:53:53 AEDT"))
Error in data.table::fifelse(length(a) == 1, a, as.POSIXct("2017-12-26 17:53:53 AEDT")) : 
  'yes' has different class than 'no'. Please make sure that both arguments have the same class.

why is it different? I thought these functions would just ignore the TRUE option when the logical test returns FALSE. Is there a solution that doesn't involve converting back to POSIXct?

Upvotes: 1

Views: 208

Answers (1)

Ian Campbell
Ian Campbell

Reputation: 24790

How about the good old fashioned way?

if(length(a)==1){a}else(as.POSIXct("2017-12-26 17:53:53 AEDT"))
[1] "2017-12-26 17:53:53 EST"

Upvotes: 1

Related Questions