On Tue, 5 Mar 2002, Juergen Hannappel wrote: > > after all this discussion, don't you think its a good idea to add msec to > > TDatime (and its now() ctor) ? > > if a finer Grain is added, add nsecs. They fit as well into 32 bits as > usecs and even if now they are not filled to their possible prcision > it might come sooner than expected... <sigh> I agree, and this is exactly what TTimeStamp does.... it adds a ns member. It also uses POSIX standard t=0 "epoch" start (Jan 1, 1970) rather than Jan 1, 1995 (TDatime). The question is how to *fill* the sub-second portion via system routines. Via user specified values isn't a problem -- lots of options are given there including passing in a timespec struct and explicitly specifying the various components. The problem is that while there are widely implemented POSIX standard means of retrieving the system clock's seconds since "epoch", there aren't standard ways that are implemented on all platforms for getting the sub-second parts. -robert Robert W. Hatcher | rhatcher@fnal.gov 630-840-3102 FNAL CD/CP (MINOS) | MS 220, PO Box 500, Batavia IL 60510
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:50:44 MET