The more, the interesting, consider: DateAndTime class>>now " [ 10000 timesRepeat: [ self now. ] ] timeToRun / 10000.0 . If calls to DateAndTime-c-#now are within a single millisecond the semaphore code to ensure that (self now <= self now) slows things down considerably by a factor of about 20. The actual speed of a single call to DateAndTime-now in milliseconds is demonstrated by the unguarded method below. [ 100000 timesRepeat: [ self todayAtMilliSeconds: (self milliSecondsSinceMidnight) ] ] timeToRun / 100000.0 . 0.00494 0.00481 0.00492 0.00495 " | nanoTicks msm | nanoTicks := (msm := self milliSecondsSinceMidnight) * 1000000. (LastTick < nanoTicks) ifTrue: [ LastTick := nanoTicks. ^ self todayAtMilliSeconds: msm]. LastTickSemaphore critical: [ LastTick := LastTick + 1. ^ self todayAtNanoSeconds: LastTick] okay, since we multiply value by 1000000 we get a nanosecond precision. Right? :) Most funny is DateAndTime class comment (see how it is coherent with TimeStamp implementation): ------------ I represent a point in UTC time as defined by ISO 8601. I have zero duration. My implementation uses three SmallIntegers and a Duration: jdn - julian day number. seconds - number of seconds since midnight. nanos - the number of nanoseconds since the second. offset - duration from UTC. The nanosecond attribute is almost always zero but it defined for full ISO compliance and is suitable for timestamping. -------------- so i really wonder if it worth having so much for so nothing? :) -- Best regards, Igor Stasenko.