This function only needs implementation if t exposed to be a polymorphic variant. Despite what the type reads, this does *not* produce a function after reading; instead it takes the constructor tag (int) before reading and reads the rest of the variant t afterwards.
Note that we expose a sexp format that is not the one exposed in Time_ns_unix. The sexp is a single atom rendered as with to_string_utc, except that all trailing zeros are trimmed, rather than trimming in groups of three.
gen_incl lower_bound upper_bound produces values between lower_bound and upper_bound, inclusive. It uses an ad hoc distribution that stresses boundary conditions more often than a uniform distribution, while still able to produce any value in the range. Raises if lower_bound > upper_bound.
gen_uniform_incl lower_bound upper_bound produces a generator for values uniformly distributed between lower_bound and upper_bound, inclusive. Raises if lower_bound > upper_bound.
val of_date_ofday : zone:Core__.Zone.t->Core__.Date0.t->Ofday.t->t
val of_date_ofday_precise :
Core__.Date0.t->Ofday.t->zone:Core__.Zone.t->[ `Once of t| `Twice of t * t| `Never of t ]
Because timezone offsets change throughout the year (clocks go forward or back) some local times can occur twice or not at all. In the case that they occur twice, this function gives `Twice with both occurrences in order; if they do not occur at all, this function gives `Never with the time at which the local clock skips over the desired time of day.
Note that this is really only intended to work with DST transitions and not unusual or dramatic changes, like the calendar change in 1752 (run "cal 9 1752" in a shell to see). In particular it makes the assumption that midnight of each day is unambiguous.
Most callers should use of_date_ofday rather than this function. In the `Twice and `Never cases, of_date_ofday will return reasonable times for most uses.
val to_date_ofday : t->zone:Core__.Zone.t->Core__.Date0.t * Ofday.t
val to_date_ofday_precise :
t->zone:Core__.Zone.t->Core__.Date0.t
* Ofday.t
* [ `Only | `Also_at of t| `Also_skipped of Core__.Date0.t * Ofday.t ]
Always returns the Date.t * Ofday.t that to_date_ofday would have returned, and in addition returns a variant indicating whether the time is associated with a time zone transition.
- `Only -> there is a one-to-one mapping between [t]'s and
[Date.t * Ofday.t] pairs
- `Also_at -> there is another [t] that maps to the same [Date.t * Ofday.t]
(this date/time pair happened twice because the clock fell back)
- `Also_skipped -> there is another [Date.t * Ofday.t] pair that never happened (due
to a jump forward) that [of_date_ofday] would map to the same
[t].
val to_date : t->zone:Core__.Zone.t->Core__.Date0.t
For performance testing only; reset_date_cache () resets an internal cache used to speed up to_date and related functions when called repeatedly on times that fall within the same day.
Unlike Time_ns, this module purposely omits max_value and min_value: 1. They produce unintuitive corner cases because most people's mental models of time do not include +/- infinity as concrete values 2. In practice, when people ask for these values, it is for questionable uses, e.g., as null values to use in place of explicit options.
val convert :
from_tz:Core__.Zone.t->to_tz:Core__.Zone.t->Core__.Date0.t->Ofday.t->Core__.Date0.t * Ofday.t
It's unspecified what happens if the given date/ofday/zone correspond to more than one date/ofday pair in the other zone.
The {to,of}_string functions in Time convert to UTC time, because a local time zone is not necessarily available. They are generous in what they will read in.
val to_string_trimmed : t->zone:Core__.Zone.t->Base.String.t
Same as to_string_abs_trimmed, except it leaves off the timezone, so won't reliably round trip.
val to_sec_string : t->zone:Core__.Zone.t->Base.String.t
Same as to_string_abs, but without milliseconds and the timezone. May raise if zone offsets move the apparent value beyond min_value_representable and max_value_representable.
val to_sec_string_with_zone : t->zone:Core__.Zone.t->Base.String.t
Same as to_sec_string but includes timezone
val of_localized_string : zone:Core__.Zone.t->Base.String.t->t
of_localized_string ~zone str read in the given string assuming that it represents a time in zone and return the appropriate Time.t
of_string_gen ~default_zone ~find_zone s attempts to parse s as a t, calling out to default_zone and find_zone as needed.
val to_string_iso8601_basic : t->zone:Core__.Zone.t->Base.String.t
to_string_iso8601_basic return a string representation of the following form: %Y-%m-%dT%H:%M:%S.%s%Z e.g. to_string_iso8601_basic ~zone:Time.Zone.utc epoch = "1970-01-01T00:00:00.000000Z"
val occurrence :
[ `First_after_or_at | `Last_before_or_at ]->t->ofday:Ofday.t->zone:Core__.Zone.t->t
occurrence side time ~ofday ~zone returns a Time.t that is the occurrence of ofday (in the given zone) that is the latest occurrence (<=) time or the earliest occurrence (>=) time, according to side.
NOTE: If the given time converted to wall clock time in the given zone is equal to ofday then the t returned will be equal to the t given.
of_string_with_utc_offset requires its input to have an explicit UTC offset, e.g. 2000-01-01 12:34:56.789012-23, or use the UTC zone, "Z", e.g. 2000-01-01 12:34:56.789012Z.
next_multiple ~base ~after ~interval returns the smallest time of the form:
time = base + k * interval
where k >= 0 and time > after. It is an error if interval <= 0.
Supplying ~can_equal_after:true allows the result to satisfy time >= after.
This function is useful for finding linear time intervals, like every 30 minutes or every 24 hours. This is different from rounding to apparent clock-face internvals, like "every hour at :00 and :30" or "every day at noon", because time zone and daylight savings transitions may cause linear intervals and apparent clock-face intervals to differ.
Time zone offsets (in the tzdata time zone database, at least) are expressed in seconds, so rounding to units of seconds or smaller is not affected by time zones. The round* functions below provide some straightforward cases. For other small units that evenly divide into a second, call with base = epoch, after as the time to round, and interval as the unit span you are rounding to.
prev_multiple ~base ~before ~interval returns the largest time of the form:
time = base + k * interval
where k >= 0 and time < before. It is an error if interval <= 0.
Supplying ~can_equal_before:true allows the result to satisfy time <= before.
This function is useful for finding linear time intervals, like every 30 minutes or every 24 hours. This is different from rounding to apparent clock-face internvals, like "every hour at :00 and :30" or "every day at noon", because time zone and daylight savings transitions may cause linear intervals and apparent clock-face intervals to differ.
Time zone offsets (in the tzdata time zone database, at least) are expressed in seconds, so rounding to units of seconds or smaller is not affected by time zones. The round* functions below provide some straightforward cases. For other small units that evenly divide into a second, call with base = epoch, after as the time to round, and interval as the unit span you are rounding to.
deprecated [since 2019-01] use [to_time_float_round_nearest] or [to_time_float_round_nearest_microsecond]
*_round_nearest vs *_round_nearest_microsecond: If you don't know that you need microsecond precision, use the *_round_nearest version. *_round_nearest_microsecond is for historical purposes.