Re: SQL code to find the longest line

On 23 Oct 2012 19:48, Erick Garske wrote:

I have a SQLPRGLE program compiled to a target release of V5R4M0 that
works correctly in V7R1M0, but throws a SQL exception when processing
the prepare statement. <<SNIP>>

SQL values after processing the prepare statement:

sqlcod = 204
sqlerm =
....5...10...15...20...25...30...35...40...45...50...55...60
' *ALL *N MEM '
' '

Does anyone know if there is an issue with the above code?

The code should work fine as coded [even if there may be better ways to effect same, and even if the RTRIM scalar may be desired over TRIM]. Somewhere the database query\SQL [the message from the spooled joblog or perhaps only a trace could clarify] is using the special value *ALL to denote that the query should be performed against all of the partitions [i.e. members]. However when an existence test was performed, the code failed to handle the *ALL as a special value; instead, effectively requesting to CHKOBJ QTEMP/AA *FILE MBR('*ALL') without validating the string for a valid member name.

IMO the opening text is unclear about when\where the program does or does not experience the error. Does mention of the TGTRLS intend to imply that using Target Release of *CURRENT will avoid the problem, or that the failure is seen only on the prior or the current release? A PTF is likely required to correct the issue, but the origin for the issue is unclear; compiler and\or run-time. FWiW there are some PTFs involving processing of ALIAS names in v7; apparently defects originating as side effect of three-part naming changes.