Accessors should be compatible with all major date time classes about lubridate HOT 18 CLOSED

In particular, this means that settors should retain the class of the original object.

from lubridate.

major date time classes so far identified: Date, POSIXct, POSIXlt, timeDate, yearmon, yearqtr, zoo, timeSeries, ts, irts, fts, its, difftime

from lubridate.

Sounds like a good list.

from lubridate.

note to self: perhaps create a is.date.object() to recognize all date formats (POSIXct, POSIXt, Date, etc.) then use as.POSIXct() to put them all into one format to operate on.

Will also have to add "+.Date" etc. to the top line

from lubridate.

How should we handle this?

This code returns the same class for a settor
"second<-" <- function(x, value) {
date <- as.POSIXlt(x) - (second(x) - value)
f <- match.fun(paste("as", class(x), sep = "."))
f(date)
}

If the class does not support seconds, then "seconds<-"(x) just returns x. Do we prefer this? Or do we want to change x to a class that supports seconds (POSIXt)? Or do we want to make this an option with no change as the default?

from lubridate.

Should ymd() recognize that 14448 is "2009-07-23" in Date format (i.e. class = "Date")?

from lubridate.

"second<-" <- function(x, value) UseMethod("second")
"second<-.default" <- function(x, value) as.POSIXlt(x) - (second(x) - value)

And no ymd() should not recognise 14448 as a date.

from lubridate.

accessors now work for chron objects

I rewrote hour(x) so it returns the hour in the time zone that x is displayed in (otherwise, hour(x) would not equal the hour you see displayed on the screen for chron objects).

I made tz(x) return "GMT" for chron objects since chrons do not support timezones, but R treats them as if they have a timezone = "GMT" by default.

from lubridate.

accessors now work for zoo objects

from lubridate.

accessors now work for zooreg, yearmon, and yearqtr objects

from lubridate.

accessors now work for timeDate objects

from lubridate.

accessors now work for xts objects

from lubridate.

accessors now work for its objects

from lubridate.

difftime is not a date time object

from lubridate.

accessors now work as much as I think they're going to with ti objects from the tis package. That package shares the same function names as ours. Also, in its documentation it states the things it tries to do don't always work (see hms() documentation).

from lubridate.

accessors now work for timeSeries objects

from lubridate.

accessors now work for fts objects

from lubridate.

ts objects do not use a recognized date format. I'm not sure if I should write code to decipher them. Or if there would be any point, because if users want that sort of specificity they should use one of the above classes and not a ts class.

from lubridate.