Daylight Savings time is FUBAR

[tidyverse]

Daylight Savings time is FUBAR

[garrettgman]

I went ahead and did this, but can undo it if necessary.

So now the user can specify how to work with DST:

date <- as.POSIXct("2009-03-08 01:59:00")
  "2009-03-08 01:59:00 CST"

relative - consistent clock time
options(DST = "relative")
date + days(1)
"2009-03-09 01:59:00 CDT"

exact - consistent time lengths
options(DST = "exact")
date + days(1)
"2009-03-09 02:59:00 CDT"

[garrettgman]

This also solves
mday(date) <- 9
date
"2009-03-09 02:59:00 CDT"
setting day value, also "set" hour value. Is this desirable? User can avoid with:
options(DST = "relative")
mday(date) <- 9
date
"2009-03-09 01:59:00 CDT"

[garrettgman]

I removed the options method. When a period is added to a date-time, R will ignore DST and just change to clock time of the date appropriately. When a duration is added to a date, R will maintain exact time intervals. So if DST occurs the resulting clock time will differ by an hour.

For one hour out of the year, this will appear not to work: when springing ahead (in the spring) the hour from 2:00am - 3:00am does not exist. R will express these clock times in the 1:00 hour. But all period math works, and once the date-time leaves this hour the clock time will appear appropriately. This is an R issue and I don't know how to change it.