To close the loop, I think I have beaten MVS and LE into submission! Not
fully tested yet ...
The short version is that the first (or only) save area back pointer must
not be zero but rather = CEECAADDSA. Also I changed a line in the code I
originally posted to
AUTO=5000, Arbitrary +
If anyone wants the long version I guess you could write me off line.
Charles
-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On
Behalf Of Charles Mills
Sent: Wednesday, August 28, 2013 8:20 AM
To: [email protected]
Subject: Re: LE U4088 Reason 63 -- no back chain
Thank you to the always-helpful Lizette!
Yes, I have "looked at" a lot of things. You know how these things are: you
head down one path, you hit some negative, so you head down some other path,
and then you hit some new negative, so perhaps you go back to the first path
...
>From the LE P/G, 3.4.4.1.1:
When TRAP(ON,SPIE) is in effect, Language Environment is notified of abends
and program interrupts. Language semantics, C/C++ signal handlers, PL/I
ON-units, and user-written condition handlers can then be invoked to handle
them. An exception to this behavior is that Language Environment cannot
handle Sx22 abends, even if TRAP(ON) is specified.
I interpret that as saying that no matter how you slice it, LE will not
handle/trap/pass/indicate operator cancels.
It turns out the documentation is not really correct. I can issue an ABEND
X'122',,,SYSTEM from my own code and LE handles it perfectly. What it cannot
do is any SDWACLUP ABENDs including operator cancel, and there appears to be
no getting around that with LE facilities.
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
