Weird problem with times.

This problem didn't begin happening until after the power blinked last night, and the box on which it's happening got abruptly shut down.

In an RPG trigger program, I have a call to the published MailDate procedure:

*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* MailDate(): Returns the current date, formatted for use * in an e-mail message. * * For example: 'Sat, 23 Oct 2004 14:42:06 -0500' *+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P MailDate B D MailDate PI 31A D CEELOCT PR opdesc D Lilian 10I 0 D Seconds 8F D Gregorian 23A D fc 12A options(*omit) D CEEUTCO PR opdesc D Hours 10I 0 D Minutes 10I 0 D Seconds 8F D CEEDATM PR opdesc D input_secs 8F const D date_format 80A const options(*varsize) D char_date 80A options(*varsize) D feedback 12A options(*omit) D rfc2822 c 'Www, DD Mmm YYYY HH:MI:SS'
D junk1 s 8F D junk2 s 10I 0 D junk3 s 23A D hours s 10I 0 D mins s 10I 0 D tz_hours s 2P 0 D tz_mins s 2P 0 D tz s 5A varying D CurTime s 8F D Temp s 25A C CALLP CEEUTCO(hours: mins: junk1) C EVAL tz_hours = %abs(hours) C EVAL tz_mins = mins C IF hours < 0 C EVAL tz = '-' C ELSE C EVAL tz = '+' C ENDIF C EVAL tz = tz + %editc(tz_hours: 'X') C + %editc(tz_mins: 'X') C CALLP CEELOCT(junk2: CurTime: junk3: *omit) C CALLP CEEDATM(CurTime: rfc2822: Temp: *omit) C RETURN Temp + ' ' + tz P E

which is then followed, a few statements later by a call to a procedure of my own, that uses a perfectly ordinary RPG TIME statement, to a Z-Timestamp field.

Everything worked perfectly last night, but as of this morning, after we recovered from the unscheduled power interruption, it started throwing CPF1850s:

To module . . . . . . . . . : QRNXTIME To procedure . . . . . . . : _QRNX_TIME_DTZ Statement . . . . . . . . . : 2 Message . . . . : Format CEE not valid

When looking at the message from option 7 in WRKACTJOB, some of the spaces surrounding "CEE" are replaced with blocks.

Running a test program that loops through an identical TIME statement 10000 times works just fine. I'm about to modify the test program so that it uses the MailDate() routine.

Any idea what could be blowing this up? I had to single-step it in the debugger just to get it to where I knew when it was throwing the CPF1850, as what it was showing in the joblog made no sense.

--
JHHL