RE: [Boost-Users] Using time_duration::rep_type::ticks_per_second

From: Jeff Garland (jeff_at_[hidden])
Date: 2003-05-18 09:37:06

> I have written a function to convert a posix_time::ptime to a windows
> FILETIME (which is in essence a 64-bit value counting 100-nanosecond
> intervals since Jan 01, 1601). 
> 
> In order to do so, I do a time difference between
> the ptime and a time setup to the FILETIME reference time. To convert
> from this time_duration to the FILETIME count, I use
> time_duration::rep_type::ticks_per_second. Since this does not seem to
> be documented I just wanted to make sure I was using it correctly (or
> if I should be using it at all). 

Yes, this is the correct way to do this.

>Here is my code (which seems to work):
> 
> const ULONGLONG nanosecPerFileTimeUnit = 100;
> 
> using namespace boost::posix_time;
> using namespace boost::gregorian;
> FILETIME toFileTime
> (
>     const ptime& time
> )
> {
>     ptime refTime(date(1601,Jan,01));
>     
> 
>     time_duration sinceRefDate = time - refTime;
> 
>     ULARGE_INTEGER durationInFileTimeUnits;
>     const ULONGLONG timeDurTicksPerSec =
> time_duration::rep_type::ticks_per_second;
> 
>     const ULONGLONG fileTimeUnitsPerTimeDurTick = 
> (nanosecPerSec/timeDurTicksPerSec) / nanosecPerFileTimeUnit;
> 
>     durationInFileTimeUnits.QuadPart =
> sinceRefDate.ticks()*fileTimeUnitsPerTimeDurTick ;
> 
>     FILETIME ft;
>     ft.dwLowDateTime = durationInFileTimeUnits.LowPart;
>     ft.dwHighDateTime = durationInFileTimeUnits.HighPart;
>     
>     return ft;
> }

On quick inspections this code looks fine.  Of course it will break if 
you ever set a date before Jan 1 1601.

Jeff