On 2/4/2007 4:53 PM, Paul Gilmartin wrote:
On Sat, 4 Nov 2006 09:16:44 -0500, Gilbert Saint-Flour wrote:
The system closes DCBs at end-of-task.
You and IBM are in agreement, at least insofar as this should happen;
they've taken APAR OA19801 for a case where it doesn't. Interesting --
the APAR seems to say that an attempt is made to flush the buffer after
the task has already terminated. This seems to leave the file at least
partially open, but closed enough that I could FREE the DDNAME. I'd
almost expect a program check from an attempt to manipulate previously
freed resources.
I think I interpret the APAR text a little differently than you do, gil.
To me it seems like the UNIX task termination code runs before the
DCB-related task termination code, and so has already deleted the
UNIX-related information that underlies the DCB, before the DCB-related
code can flush the buffers and close the DCB. Thus the 614 abend that
the APAR talks about. So, the system is closing the DCB during task
termination, but we have the situation where two things need to happen
in a particular order, and they're out of order right now.
Walt
----------------------------------------------------------------------
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
