On Tue, 1 Apr 2014 18:35:07 -0400, Shmuel Metz (Seymour J.) wrote:
> on 03/31/2014 at 04:39 PM, Paul Gilmartin said:
>
>>FSVO "there" (perhaps John trimmed it). HLASM manages to display
>>PTF level on the first page of every SYSPRINT.
>
>It manages to display something, but unless the dependency graph is
>linear you can't tell what other service is on.
>
So, I asked on ASSEMBLER_LIST and got an authoritative answer. Reposting
with the writer's permission:
On 2014-04-01, at 02:46, Jonathan Scott wrote:
> Ref: Your note of Mon, 31 Mar 2014 23:26:21 -0600
>
> Paul Gilmartin <paulgboul...@aim.com> writes:
>> (Motivated by a thread in IBM-MAIN.)
>>
>> I notice that the HLASM PTF level appears on the first page of
>> every SYSPRINT.
>>
>> Does this imply that there is a unique CSECT updated by every PTF?
> Module ASMA9Z contains the PTF level. The source is not updated, but
> it is recompiled to pick up the PTF level via SYSPARM.
>
>> Implying that each PTF has its immediate predecessor as a PRErequisite?
> Yes.
>
>> Is that PTF level available as a preset GBLC symbol?
> No.
>
> The change history information shown by the INFO option (assembled
> into module ASMAINFO) is also updated for every APAR.
>
> Jonathan Scott
> IBM Hursley, UK
It's linear.
I'll answer Jon Perryman's argument with an ISV's bias:
I imagine a dialog, starting realistically:
Customer: I'm experiencing a problem with symptoms ...; test case
supplied.
Support: We have tried your test case on a system at the most current
service level. We can not reproduce the problem.
Customer: That's right. On our test system at current service level, the
problem does not occur. But it occurs on our production system at
service configuration [ ... ].
Support: Does it hurt when you do that?
...
Consider the set of PTFs as a directed connected graph, with the base
FUNCTION as a root and the PRE and REQ relations as arcs. The set
of valid service configurations is the set of likewise directed, connected,
and rooted subgraphs. I suspect the cardinality increases exponentially
with the number of PTFs. Enumerating those configurations is an N-P
problem.
The symptoms of the customer's problem may not appear in the text of
any APAR; the problem may have been resolved serendipitously, collateral
to repairing an apparently unrelated bug.
The customer wants the minimal service configuration which is a superset
of his current configuration and repairs the bug. How do you find it by
trial and error? If the dependency graph is linear, a binary search suffices.
Worst case? Granted. But remember Murphy.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
