zutempo

Overview

Dates and times are hard to get right. A day is not always 24 hours, one month after 31 January is not a date, a clock time can happen twice or not at all, and the same moment is a different date in Tokyo than in New York. Most code gets these cases right by accident, or not at all.

Temporal is JavaScript’s answer: a date and time API designed by the committee that standardises the language, specified to the last nanosecond and tested by thousands of conformance tests. zutempo brings Temporal to R.

• The right type for each job. A plain_date() is a date on a calendar, a plain_time() a time on a wall clock, an instant() an exact moment, a zoned_date_time() a moment in a time zone, and a duration() an amount of time. Each does what it says, and never silently becomes another.

• Arithmetic you can trust. Adding a month, a day or 24 hours each mean what they should, across month ends, leap years and clock changes, and the ambiguous cases ask you how to resolve them instead of guessing.

• The same answers everywhere. zutempo follows the Temporal specification operation by operation, so a result in R is the result in a browser or in Node. It is checked against test262, the specification’s own conformance suite.

• At home in R. Values are vectors with NA, fit in data frames, convert to and from Date and POSIXct, and read time zones from the same database base R uses. The engine is in C, with no runtime dependencies.

Installation

zutempo is not on CRAN yet. You can install the development version from GitHub:

# install.packages("pak")
pak::pak("pedrobtz/zutempo")

Usage

library(zutempo)

A team meets every day at 9:30 in New York. Clocks go forward on the night of 7 March, so the next meeting is one day later, but only 23 hours later:

meeting <- zoned_date_time("2026-03-07T09:30[America/New_York]")
temporal_add(meeting, "P1D")
#> <zutempo_zoned_date_time[1]>
#> [1] 2026-03-08T09:30:00-04:00[America/New_York]
temporal_until(meeting, temporal_add(meeting, "P1D"))
#> <zutempo_duration[1]>
#> [1] PT23H

Where is everyone else at that moment?

temporal_with_time_zone(meeting, c("Europe/Lisbon", "Asia/Tokyo"))
#> <zutempo_zoned_date_time[2]>
#> [1] 2026-03-07T14:30:00+00:00[Europe/Lisbon]
#> [2] 2026-03-07T23:30:00+09:00[Asia/Tokyo]

Calendar arithmetic knows that months have different lengths:

due <- plain_date(c("2026-01-31", "2026-03-31", "2026-05-15"))
temporal_add(due, "P1M")
#> <zutempo_plain_date[3]>
#> [1] 2026-02-28 2026-04-30 2026-06-15
temporal_until(plain_date("2026-01-31"), "2026-12-25", largest_unit = "month")
#> <zutempo_duration[1]>
#> [1] P10M25D

And durations are values you can round, compare and total:

temporal_round(duration(minutes = 200), largest_unit = "hour")
#> <zutempo_duration[1]>
#> [1] PT3H20M
duration("PT90M") < duration("PT2H")
#> [1] TRUE
temporal_total(duration("P1M"), "day", relative_to = "2026-02-01")
#> [1] 28

Every function follows one naming rule: a Temporal type is a constructor (plain_date()), a method is temporal_ and its name (temporal_add()), and a property is a function of its name (year(), hours_in_day()). If you know Temporal, you already know zutempo.

Learn more

• Get started follows one value from construction to base R.
• Articles on plain dates and times, durations, instants and time zones, strings and formatting, and base R and data frames.
• The reference lists every Temporal member and the R function that implements it.

Acknowledgements

zutempo is an implementation of the work of the TC39 Temporal champions, who designed the API and wrote its specification and its test262 tests. temporal-polyfill is the reference zutempo is checked against, and time zones come from the IANA time zone database. lubridate and clock showed how much R users care about getting dates and times right.