Dynamically Change as.POSIXlt Value - r

Question

In R, I am trying to read a file that has a timestamp, and update the timestamp based on the condition of another field. The below code works with no problem:
t <- data.frame(user = as.character(c("bshelton#email1.com", "lwong#email1.com")),
last_update = rep(as.POSIXlt(Sys.time(), tz = "America/Los_Angeles"), 2))
Sys.sleep(5)
t$last_update <- as.POSIXlt(ifelse(t$user == "bshelton#email1.com", Sys.time(), t$last_update), origin = "1970-01-01")
print(t)
The problem is when I read an existing file and try to dynamically change an as.POSIXlt value. The following code is producing the error that accompanies it in the code block afterwards:
t <- data.frame(user = as.character(c("bshelton#email1.com", "lwong2#email1.com")),
last_update = rep(as.POSIXlt(Sys.time(), tz = "America/Los_Angeles"), 2))
write.csv(t, "so_question.csv", row.names = FALSE)
t <- read.csv("so_question.csv")
t$last_update <- as.POSIXlt(t$last_update)
Sys.sleep(5)
t$last_update <- as.POSIXlt(ifelse(t$user == "bshelton#email1.com", Sys.time(), t$last_update), origin = "1970-01-01")
Error in as.POSIXlt.default(ifelse(t$user == "bshelton#email1.com", Sys.time(), :
do not know how to convert 'ifelse(t$user == "bshelton#email1.com", Sys.time(), t$last_update)' to class “POSIXlt”
In addition: Warning message:
In ans[!test & ok] <- rep(no, length.out = length(ans))[!test & :
number of items to replace is not a multiple of replacement length

Answer 1

The first case is curiously working only because you don't have what you think—those datetimes are in fact POSIXct, not POSIXlt:
last_update <- rep(as.POSIXlt(Sys.time(), tz = "America/Los_Angeles"), 2)
str(last_update)
#> POSIXlt[1:2], format: "2019-07-28 20:52:10" "2019-07-28 20:52:10"
t <- data.frame(user = as.character(c("bshelton#email1.com", "lwong#email1.com")),
last_update = last_update)
str(t)
#> 'data.frame': 2 obs. of 2 variables:
#> $ user : Factor w/ 2 levels "bshelton#email1.com",..: 1 2
#> $ last_update: POSIXct, format: "2019-07-28 20:52:10" "2019-07-28 20:52:10"
If you dig into ?data.frame, it says
data.frame converts each of its arguments to a data frame by calling as.data.frame(optional = TRUE). As that is a generic function, methods can be written to change the behaviour of arguments according to their classes: R comes with many such methods. Character variables passed to data.frame are converted to factor columns unless protected by I or argument stringsAsFactors is false. If a list or data frame or matrix is passed to data.frame it is as if each component or column had been passed as a separate argument (except for matrices protected by I).
This is what's happening: as.data.frame.POSIXlt in fact converts to POSIXct:
now <- Sys.time()
str(now)
#> POSIXct[1:1], format: "2019-07-28 22:50:12"
str(data.frame(time = now))
#> 'data.frame': 1 obs. of 1 variable:
#> $ time: POSIXct, format: "2019-07-28 22:50:12"
as.data.frame.POSIXlt
#> function (x, row.names = NULL, optional = FALSE, ...)
#> {
#> value <- as.data.frame.POSIXct(as.POSIXct(x), row.names,
#> optional, ...)
#> if (!optional)
#> names(value) <- deparse(substitute(x))[[1L]]
#> value
#> }
#> <bytecode: 0x7fc938a11060>
#> <environment: namespace:base>
More immediately, since Sys.time() returns a POSIXct object, ifelse(t$user == "bshelton#email1.com", Sys.time(), t$last_update) in the second case is getting a POSIXct object for one observation and POSIXlt for the other. The POSIXlt object's class attribute is dropped by ifelse revealing the list underneath, which ifelse then doesn't know how to turn into a vector together with the unclassed POSIXct object (which is just a number).
The solution here, then, is to follow the hint data.frame is giving you and use POSIXct instead of POSIXlt.
If you really want to make it work with POSIXlt, you can iterate over the conditions and POSIXlt vector with Map with if/else (which maintain attributes including class, but only handle scalar conditions) and coerce the resulting list back to a vector with do.call(c, ...):
t <- data.frame(user = as.character(c("bshelton#email1.com", "lwong#email1.com")),
last_update = rep(as.POSIXlt(Sys.time(), tz = "America/Los_Angeles"), 2))
t$last_update <- as.POSIXlt(t$last_update)
t$last_update <- do.call(c, Map(
function(condition, last_update){
if (condition) {
as.POSIXlt(Sys.time() + 5)
} else {
last_update
}
},
condition = t$user == "bshelton#email1.com",
last_update = t$last_update
))
t
#> user last_update
#> 1 bshelton#email1.com 2019-07-28 23:11:04
#> 2 lwong#email1.com 2019-07-28 23:10:59
...but frankly that's a little silly. Just use POSIXct instead, and your life will be better.