silly $GETQUI bug and work-around

[Volker Halle]

Jess,

what kind of contest is this ? Are we going to do quality testing for HP OpenVMS engineering here ? Believe, me, I'm up to that ;-)

When converting 20-NOV-2006 22:02:14.40 to a quadword time value, you might note, that the low order longword becomes all zero.

The code in [CLIUTL]QUEMANSHO tests whether a RETENTION_TIME is specified, but only tests the low order longword of the quadword time value.

You've found the symptom of this bug, I found the actual bug in the source code. So 90% of the work is done. Now it would just take someone with a support contract or access to PTR to report this problem officially and wait for OpenVMS engineering to fix it.

Volker.

[John Abbott_2]

> Now it would just take someone with a support contract or access to PTR to report this problem officially and wait for OpenVMS engineering to fix it.

If you *REALLY* think it's worth it (maybe because most of the work is already done by Volker) then I'm sure someone will log a support call (maybe me) or engineering might take note (ye olde Mr DCL did!)

Just post exactly what needs to be said...

J.

[Jess Goodman]

I did send an email mentioning these two bugs and got a polite reply from David Sweeney. I hope fixing them doesn't take time away from anything important.

Volker, no this second bug is not that simple. It does not occur for all binary time values with a zero low-order long word:

DBG> exam/hex r0
%R0: 00A5F20000000000
DBG> exam/date r0
%R0: 22-NOV-2006 20:48:17.11

$ SUBMIT JUNK /RETAIN=UNTIL="22-NOV-2006 20:48:17.11"
Job JUNK (queue SYS$BATCH, entry 5009500) pending

$ SHOW ENTRY/FULL '$ENTRY
Entry Jobname Username Blocks Status
----- ------- -------- ------ ------
5009500 JUNK SYSTEM Pending (queue stopped)
On stopped batch queue SYS$BATCH
Submitted 8-NOV-2006 19:29:14.17 /PRIORITY=100
/RETAIN=UNTIL="22-NOV-2006 20:48"
File: _$1$DGA111:[DSKA.COM.SYSTEM]JUNK.COM;3

If it was based on a zero low-order long word the bug would occur for any value that is a multple of delta time interval of 0-07:09.49 (binary time value of negative 2**32). As stated above the delta interval for this problem is actually 15-12:49:37.28, whose binary value is negative (2**32)x3125.

[Volker Halle]

Jess,

I give up for now. To troubleshoot this effectively, one would need to run the code with the debugger. Let's leave this to OpenVMS engineering...

Volker.

[Volker Halle]

Jess,

both problems have now been logged via PTR (official OpenVMS problem tracking tool) at HP:

75-13-1804 for the /retain=until problem
75-13-1805 for the cpu_limit problems

You may consider to close this thread.

Volker.

[Jess Goodman]

thread closed