The obvious problem here is that the manuals specify that SYSLRACF
returns the RACF level [as a character string], but give zero guidance
about valid ways to interpret or use the value.  If this will always be
the suffix of an FMID AND the rules for RACF FMID assignment guarantee
that comparing as character-encoded hexadecimal values will always sort
later releases higher, then that ought to have been explicitly stated.

The problem with unfreezing SYSLRACF is the possibility of breaking
existing code using other comparison techniques that worked prior to
level "77A0".  It would seem that there are several plausible choices at
this point:
(1) Fix the documentation to be specific about valid usage of SYSLRACF
and warn of future "unfreeze" of SYSLRACF to reflect true FMID, fix any
mis-use in IBM code, and require installations to fix any installation
misuse as part of some future migration.
(2) Keep SYSLRACF as-is and introduce a new SYSVAR for true RACF FMID
suffix with well defined usage rules, and require any new code needing
to verify RACF at a level of "77A0" or above to use the new variable and
the explicitly-stated comparison rules.
      or,
if there is no rule that FMID suffixes taken as a hex value always
increase in later product releases,
(3) introduce a new SYSVAR that returns some kind of RACF level set #
guaranteed to increase sequentially for later releases and  require that
future code compare with that for determining availability of features
at "77A0" and beyond.  It could still be useful to have a SYSVAR with
the true RACF FMID suffix, even if that is not the best thing to test to
determine available features.

    Joel C. Ewing


On 6/24/21 2:38 PM, Paul Gilmartin wrote:
> On Mon, 21 Jun 2021 14:35:46 -0700, Charles Mills  wrote:
>
>> Did you read the doc? They are concerned because 77A0 will character compare 
>> low to 7790 and mess up peoples' logic. Seems to me if you do character 
>> compares on hex data you get what you deserve, but I don't make up the rules.
>>
> Errr ...
> "77A0" < "7790" (EBCDIC)
> "77A0" > "7790" (ASCII)
>
> Do they need to specify the CCSID?  (I believe DFSORT provides such an 
> option.)
>
> The doc is probably ASCII-centric.
>
> -- gil
>
> ...


-- 
Joel C. Ewing

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

Reply via email to