na.locf.xts() wrong when only first row of multi-column object is not NA

[joshuaulrich]

Calling na.locf() on an xts object that is all NA except for the first row doesn't propagate the first row value to all subsequent rows. Coercing to zoo and running na.locf() works correctly and caries the values forward.

x <- .xts(matrix(NA_integer_, 5, 2), 1:5)
x[1,] <- 1:2
na.locf(x)
#                     [,1] [,2]
# 1969-12-31 18:00:01    1   NA
# 1969-12-31 18:00:02    1   NA
# 1969-12-31 18:00:03   NA   NA
# 1969-12-31 18:00:04   NA   NA
# 1969-12-31 18:00:05   NA   NA
na.locf(as.zoo(x))
#                        
# 1969-12-31 18:00:01 1 2
# 1969-12-31 18:00:02 1 2
# 1969-12-31 18:00:03 1 2
# 1969-12-31 18:00:04 1 2
# 1969-12-31 18:00:05 1 2

Thanks to Christopher Katsulis (via email) for the report and reproducible example!

[ckatsulis]

In windows, I am currently getting the following error

Error in na.locf.xts(data) : 
  na.locf.xts only handles univariate, dimensioned data

[joshuaulrich]

In windows, I am currently getting the following error

Error in na.locf.xts(data) :
na.locf.xts only handles univariate, dimensioned data

Following up here, since this was resolved via private email. Here's my reply:

That's weird. It suggests you're not using the latest master from GitHub, because I don't see that error text in the code:

josh@thinkpad: /storage/git/R/xts (master)
> grep "only handles" -r *

josh@thinkpad: /storage/git/R/xts (master)

Can you double-check that you have the latest HEAD from master?

And I asked in a different email thread:

Did you get your Windows na.locf.xts "by-column" error sorted out?

and Chris replied

I believe yes.... deleted the folder and reinstalled from github