On Thu, 17 Nov 2016 10:46:37 -0500, Kurt Quackenbush wrote:

>> I don't believe PTFS using IEBUPDTE can be RESTORED.
>
>Why do you believe this?  Of course PTFs containing ++MACUPD or ++SRCUPD
>can be RESTOREd.
> 
I stand corrected.  I misled myself by my (mis-)understanding of our own
development process.  Several developers perform integration testing in
a common, very volatile CSI.  SYSMODs are ACCEPTed only at the point
at which they are passed downstream to EVT.  I've learned several things
to my dismay:

o APPLY REDO doesn't always undo the side effects of the prior APPLY.
  I suspect this is particularly true if the SYSMOD contains ++*UPD or
  ++ZAP or if the newer version contains fewer CSECTs than the
  previous and this is what I misremember as "can't RESTORE".

o RECEIVE with a higher REWORK value doesn't always undo the side
  effects of the previous RECEIVE.  Our scripts always REJECT before
  RECEIVE and ignore failure.  (Is there a better way to test whether a
  SYSMOD has previously been RECEIVEd and bypass the REJECT if
  it has not?)

On a very few occasions, a CSI has gotten so damaged that I have needed
to reinstall the base and all current service.  Our developers cringe:  "How
would this play in a customer shop?"  It's no problem if the offending PTF
has not been released to field.

Such complexities impel my wish that LINK LMODS could appear in MCS.
And that a SYSMOD containing LINK LMODS could be RESTOREd.
(Grass is always greener ...?  -- Colorado joke.)

How do my colleagues deal with such things?  One CSI per developer isn't
prohibitive, but often those developers want to test SYSMOD interactions.

But as SMP/E supports source updates, it's disappointing that the more
flexible patch(1) isn't available as an alternative to IEBUPDTE.  And that
APPLY REDO doesn't reverse the (putative) prior patch.

--gil

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

Reply via email to