[R-Forge #1603] Can't use na.locf() on character data

[joshuaulrich]

Submitted by: Ken Williams (@kenahoo)
Assigned to: Nobody
R-Forge link

na.locf() seems to fail on xts objects containing character data:

s1 <- xts(c('OK', 'Fail', NA, 'Service'), Sys.Date()+1:4)
na.locf(s1)
Error in na.locf.xts(s1) : unsupported type

s1 <- xts(c(1,2,NA,4), Sys.Date()+1:4)
na.locf(s1)
[,1]
2011-09-28 1
2011-09-29 2
2011-09-30 2
2011-10-01 4

I can use na.locf.default() as a workaround, I don't seem to lose any fidelity in the process:

s1 <- xts(c('OK', 'Fail', NA, 'Service'), Sys.Date()+1:4)
na.locf.default(s1)
[,1]
2011-09-28 'OK'
2011-09-29 'Fail'
2011-09-30 'Fail'
2011-10-01 'Service'

-Ken

[SamoPP]

Also seems that na.omit fails on all character data:

> na.omit(xts(c("a", "b", "c", NA, "d"), as.Date(1:5)))
Error in na.omit.xts(xts(c("a", "b", "c", NA, "d"), as.Date(1:5))) : 
  unsupported type

[Masutani]

xout option for na.locf / na.locf.default cannot work for character data (xts/zoo) .
Failed with "zero non-NA points" error in approx .

I found the process is fine until reaching to 26 line in na.approx
objectm <- merge(object.x, zoo(, xout))
https://github.com/cran/zoo/blob/master/R/na.approx.R

[joshuaulrich]

@kenahoo, I hope "better late than never" applies in this case!

I didn't address the n-way merge issue on character xts because I have some significant changes to the merge C code that I would like to incorporate first.