Documentation ¶
Index ¶
Constants ¶
const ClocklessMaxOffset = math.MaxInt64
ClocklessMaxOffset is a special-cased value that is used when the cluster runs in "clockless" mode. In that (experimental) mode, we operate without assuming any bound on the clock drift.
Variables ¶
var UnixEpoch = time.Unix(0, 0)
UnixEpoch represents the Unix epoch, January 1, 1970 UTC.
Functions ¶
func FromUnixMicros ¶
FromUnixMicros returns the local time.Time corresponding to the given Unix time, usec microseconds since UnixEpoch. In Go's current time.Time implementation, all possible values for us can be represented as a time.Time.
func LoadLocation ¶
LoadLocation returns the time.Location with the given name. The name is taken to be a location name corresponding to a file in the IANA Time Zone database, such as "America/New_York".
We do not use Go's time.LoadLocation() directly because: 1) it maps "Local" to the local time zone, whereas we want UTC. 2) when a tz is not found, it reports some garbage message related to zoneinfo.zip, which we don't ship, instead of a more useful message like "the tz file with such name is not present in one of the standard tz locations".
func Now ¶
Now returns the current local time with an optional offset specified by the environment. The offset functionality is guarded by the "clockoffset" build tag - if built with that tag, the clock offset is parsed from the "COCKROACH_SIMULATED_OFFSET" environment variable using time.ParseDuration, which supports quasi-human values like "1h" or "1m".
func ToUnixMicros ¶
ToUnixMicros returns t as the number of microseconds elapsed since UnixEpoch. Fractional microseconds are rounded, half up, using time.Round. Similar to time.Time.UnixNano, the result is undefined if the Unix time in microseconds cannot be represented by an int64.
Types ¶
type Timer ¶
type Timer struct { // C is a local "copy" of timer.C that can be used in a select case before // the timer has been initialized (via Reset). C <-chan time.Time Read bool // contains filtered or unexported fields }
The Timer type represents a single event. When the Timer expires, the current time will be sent on Timer.C.
This timer implementation is an abstraction around the standard library's time.Timer that provides a temporary workaround for the issue described in https://github.com/golang/go/issues/14038. As such, this timer should only be used when Reset is planned to be called continually in a loop. For this Reset pattern to work, Timer.Read must be set to true whenever a timestamp is read from the Timer.C channel. If Timer.Read is not set to true when the channel is read from, the next call to Timer.Reset will deadlock. This pattern looks something like:
var timer timeutil.Timer defer timer.Stop() for { timer.Reset(wait) switch { case <-timer.C: timer.Read = true ... } }
Note that unlike the standard library's Timer type, this Timer will not begin counting down until Reset is called for the first time, as there is no constructor function.
func (*Timer) Reset ¶
Reset changes the timer to expire after duration d and returns the new value of the timer. This method includes the fix proposed in https://github.com/golang/go/issues/11513#issuecomment-157062583, but requires users of Timer to set Timer.Read to true whenever they successfully read from the Timer's channel.
func (*Timer) Stop ¶
Stop prevents the Timer from firing. It returns true if the call stops the timer, false if the timer has already expired, been stopped previously, or had never been initialized with a call to Timer.Reset. Stop does not close the channel, to prevent a read from succeeding incorrectly.