> When you pass this field to the intrinsic "HPDATECONVERT" it will convert
> the longint-type to a real date such as 20050207 for Feb 7, 2005. Fine. I am

Uh, you mean "human readable" date? :)  the microseconds since 1970
is just as "real" a date as a string of 8 digits!

> So, can anyone out there tell me how to convert a real date AND time
> sequence (e.g. 20050307 12:34:56) to "microseconds since January 1, 1970"
> for the impending AIF call to update the timestamps in a file?

The basic answer is:   +

I'll expand on that...

Summary of problem:
   You're trying to convert a yyyymmdd hh:mm:ss date/time into
   microseconds-since-1970, and you've just discovered an omission
   in the date/time formats that HPDATECONVERT supports.

Background:
   Oops, my fault ... I should have thought of that when I proposed
   that intrinsic!  In retrospect, there should have been at least
   one more format supported by the intrinsic, an 8 byte structure
   with a 4 byte CALENDAR value followed by a 4 byte CLOCK value.

Solution:

   1) If COBOL can't do 64-bit integer arithmetic:

      a) convert the yyyymmdd to type 10 (int32, seconds since 1970)

      b) add to that:

                 hh * 3600 + mm * 60 + ss

      c) HPDATECONVERT the sum (which is type 10 date/time) to type 1
         (int64, microsecs since 1970)

   2) if COBOL can do 64-bit integer arithmetic:

      a) convert the yyyymmdd to type 10 (int32, seconds since 1970)

      b) add to that:

                 hh * 3600 + mm * 60 + ss

      c) multiply by one million

Stan

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *