Mr Butz writes:
| I am just subtracting the before and after microseconds portion ignoring date
and time
Consider a four-position decimal counter that currently has the value 99, i.e.,
| 0 | 0 | 9 | 9 |
Now save the low-order two digits, 99, and add 7 to the counter, obtaining
| 0 | 1 | 0 | 6 |
Saving the low-order two digits again yields 06. Then the subtraction 06 - 99
yields a difference of -07, i.e., nonsense.
Use the STCKF instruction, which 1) is a faster, lower-overhead one and 2)
stores binary zeros to the right of the position that is being incremented.
Shift your STCKF values right to eliminate these insignificant binary zeros.
Then consult your PrOp again with some care, reconsidering how properly to
interpret differences in these shifted values.
The notion that an STCKx value has separate date, time (hours?), and
microsecond fields is just wrong. It is a counter that has, conceptually, been
incremented by one for each elapsed clock unit since midnight 1899 December 31.
You may discard precision on the right; but you cannot ignore leftmost,
higher-order bits in any circumstances.
John Gilmore Ashland, MA 01721-1817 USA
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
