Re: Need a small VMS test

On Sep 30, 2008, at 6:22 AM, Hein, Nashua NH wrote:

> On Sep 30, 1:24 am, [email protected] (John E. Malmberg) wrote:
>> Michael G Schwern wrote:
>>> Hi, I need a quick VMS test for the new code to detect the range  
>>> of time_t.
>
>> VMS keeps time internally as a 64 bit signed integer all platforms,
>> usually in local time.  The base time is November 17, 1958.
>
> Close. They are 100 ns granularity cluncks with a signed 64 bit
> counter since the Smithsonian base date 00:00 17-Nov-1858.
>
> So it will happily represent the date of birth for a centenarian, as
> well as the end date for a new 99 year lease.
>
> The build-in ascii formatting is restricted to the year 9999 and
> typically includes centi-seconds.
>
> The negative times are considered 'delta times', which are formate to
> DDDD HH:MM:SS.CC

Nice summary.  None of which has anything to do (directly) with the  
question, which was about the behavior of time_t.  As John's test  
showed, time_t is by default a 32-bit unsigned value, but can be  
forced to be signed with a compiler flag.  There are some indications  
in the headers that 64-bit time_t will be supported in the future.  If  
and when that happens, it's impossible to predict whether 64-bit will  
ever be the default or whether it will be signed or unsigned.
________________________________________
Craig A. Berry
mailto:[email protected]

"... getting out of a sonnet is much more
  difficult than getting in."
                  Brad Leithauser