31 May
2014
31 May
'14
12:26 p.m.
Hi, Alexander! On May 29, Alexander Barkov wrote:
I still think that we should rely only on the range and should not check the fractional part.
...
If the non-zero fractional part fits into 6 digits, it considers the fractional part as microseconds.
If the leading 6 fractional digits are zeros and there are some non-zeros in the 7th position, it works differently and obviously breaks the rule "treat as TIME if there is a fractional part".
With floating point numbers it gets even more confusing to rely on the fractional part, because "fractional part is zero" is a very non-precise fact.
Okay, let's change it :) Regards, Sergei