🚀 go-pugleaf

I gave up on TEST when LE arrived, since we didn't get whatever the follow
on product to COBTEST was. IIRC COBTEST and PLITEST were part and parcel
of VS COBOL II and PL/I v2r3, but the follow on product was a separate
monthly charge item.

I was asked to help someone locate the source of an S0C7 abend yesterday.

Staring at the CEEDUMP I noticed an information message saying that it
hadn't tried to display variable values because it wasn't compiled with
TEST(SYM). There were also some lables indicating where statement numbers
might be found, but they were all empty.

Recompiled with TEST(NONE,SYM) and CEEDUMP started to display variables and
statement numbers. The compile listing also showed a new third TEST parm,
NOSEPARATE.

None of this is documented in the v2r1 manuals, even the web bookserver 
versions. It does seem to be fully documented in the v2r2 programming guide.

The SEPARATE option seems like a way of avoiding virtual storage headaches
in crowed regions such as IMS Message Processing Regions or CICS Regions
if TEST(SYM) bloats the load modules too much. 

They also avoided the possibility of a loop when printing variable contents
after a S0C7. The variable with bad decimal data was printed as character
data, so trying to print it didn't generate a second S0C7 for the same abend.

Looks like IBM is giving Abend-Aid a serious run for it's money. Abend-Aid
will use PL/I GOSTMT information to print statement numbers on the ABENDAID
report, but so far it doesn't seem to be picking up on this new source of
information. TEST(SYM/NOSYM) isn't on the list of recommended compile
options.
--