Library

Module

Module type

Parameter

Class

Class type

POSIX time spans.

**WARNING.** A POSIX time span is not equal to an SI second based time span see the basics.

## POSIX time spans

`type t = span`

The type for signed, picosecond precision, POSIX time spans.

`val v : (int * int64) -> span`

`val of_d_ps : (int * int64) -> span option`

`of_d_ps (d, ps)`

is a span for the signed POSIX picosecond span `d`

* 86_400e12 + `ps`

. `d`

is a signed number of POSIX days and `ps`

a number of picoseconds in the range [`0`

;`86_399_999_999_999_999L`

]. `None`

is returned if `ps`

is not in the right range.

`val to_d_ps : span -> int * int64`

`to_d_ps d`

is the span `d`

as a pair `(d, ps)`

expressing the POSIX picosecond span `d`

* 86_400e12 + `ps`

with `ps`

in the range [`0`

;`86_399_999_999_999_999L`

]

`val of_int_s : int -> span`

`of_int_s secs`

is a span from the signed integer POSIX second span `secs`

.

`val to_int_s : span -> int option`

`to_int_s d`

is the span `d`

as a signed integer POSIX second span, if `int`

's range can represent it (note that this depends on `Sys.word_size`

). Subsecond precision numbers are truncated.

`val of_float_s : float -> span option`

`of_float_s secs`

is a span from the signed floating point POSIX second span `d`

. Subpicosecond precision numbers are truncated.

`None`

is returned if `secs`

cannot be represented as a span. This occurs on `Stdlib.nan`

or if the duration in POSIX days cannot fit on an `int`

(on 32-bit platforms this means the absolute magnitude of the duration is greater than ~2'941'758 years).

`val to_float_s : span -> float`

`to_float_s s`

is the span `d`

as floating point POSIX seconds.

**Warning.** The magnitude of `s`

may not be represented exactly by the floating point value.

## Predicates

`compare d d'`

is a total order on durations that is compatible with signed time span order.

## Arithmetic

**Note.** The following functions rollover on overflows.

## Rounding

`round ~frac_s t`

is `t`

rounded to the `frac_s`

decimal fractional second. Ties are rounded away from zero. `frac_s`

is clipped to the range [`0`

;`12`

].

`truncate ~frac_s t`

is `t`

truncated to the `frac_s`

decimal fractional second. `frac_s`

is clipped to the range [`0`

;`12`

].

## Pretty printing

`val pp : Format.formatter -> span -> unit`

`pp ppf d`

prints an unspecified, approximative, representation of `d`

on `ppf`

.

The representation is not fixed-width, depends on the magnitude of `d`

and uses locale independent SI prefixes on seconds and accepted non-SI units. Years are counted in Julian years (365.25 SI-accepted days) as defined by the International Astronomical Union (IUA).

The representation is approximative. In particular beyond 60 seconds it only keeps the two most significant time units and rounds towards the infinity. The latter means that case arising, it always *over* approximates durations.

**Warning** Becomes unprecise (but does not overflow) if the absolute number of POSIX days in the time span is greater than ```
max_int /
4
```

(on 32-bit platforms this is ~735'439 years)

`val dump : Format.formatter -> span -> unit`

`dump ppf s`

prints an unspecified raw representation of `d`

on `ppf`

.