<snip>
IMHO, we need an enhancement to CSVQUERY/CSVINFO (as appropriate) to 
return the fully-qualified data set name and volume and/or HFS path from 
which a module was actually fetched. (If it came from VLF, that 
information would need to be preserved at the time the module is cached 
so it can be provided to CSV.)
</snip>

I don't see this ever happening. The system does not keep that 
information. The extra cycles to determine that information cannot be 
justified at this time. The system also does not keep information about 
whether the fetch was satisfied from LPA, LNKLST, joblib, steplib, some 
tasklib or some user-specified DCB.

By the way, CSV never actually knows the full file system path name. It 
knows only when USS provides it (which is not necessarily the full name). 
CSVQUERY and CSVINFO *do* provide the file system path name to the 
invoker.

<snip>
>However, it's not trivial to determine from where you were loaded. It
>could be STEPLIB/JOBLIB, it could be LPA, it could be LNKLST. 

It shouldn't be that hard if you know the member name. Create a DCB 
for STEPLIB and open it. If that works, do a BLDL on the member name 
and if that works, you've found the module. If the BLDL fails, it's not in 

STEPLIB and JOBLIB isn't used. If the open fails, try the same with 
JOBLIB.
</snip>

It's not "hard" but it's not trivial either.

If you're running under the job, there's no reason to create a DCB for 
STEPLIB and open it, as the DCB already exists and is open. You could just 
do a BLDL yourself using the DCB pointed to by TCBJLB.

Ignoring the user-specified DCB case, the system search is, approximately:
Do for every task from this task repeating using TCBOTC up through the 
jobstep program task
  look for the member in the TCBJLB task if there is one
End Do
Look for the member in LPA
Look for the member in LNKLST

What the system has, and could return (indeed does provide to the CSVFETCH 
exit as of z/OS 2.2) is the UCB address and CCHH of the data set. I don't 
claim to know exactly how, but you can get from that to the data set name. 
An enhancement could be made to CSVQUERY/CSVINFO to provide that data.

Peter Relson
z/OS Core Technology Design


----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to