On Sat, 16 Oct 2010 18:56:23 -0500, Joel C. Ewing wrote: > >If you were very unlucky, the allocated extents could contain data with >compatible block structure and possibly even records that could pass >application integrity checks. Subsequent programs could use the "bogus" >records to perform unintended updates and other unintended actions. I >can remember at least one or two instances where records were >successfully read from an "empty" file. > >"Unexpected program behavior" is too benign a description, because in >the latter case a missing EOF could lead to data loss, data corruption, >or other incorrect program results and there might not even be any >obvious program failure to make it evident you had a data integrity issue. > "Integrity" has a conventional meaning mindful of IBM's Integrity Statement. It's unduly alarmist here. A large class of programming errors or oversights can lead to data loss. This situation is very similar to use of an area uninitialized after GETMAIN. And users experienced "Unexpected program behavior" when z/OS optimized memory management with the effect of leaving residual data in areas that programmers had previously, unwittingly, counted on to be cleared. IBM provided a compatibility option to restore the earlier behavior. Did IBM call this a "Data Integrity" facility and recommend that customers operate in old mode? (BTW, I'm curious: what fraction of sites continue to operate in old mode?)
That said, I welcome the change to allocation; I even feel it didn't go far enough. Simply, allocation should write an end of file at the beginning of every new data set, SMS or non-SMS, and regardless of DSORG, known or unknown. The overhead would be minor compared to the uncertainty removed. Decades ago I encountered this "Unexpected program behavior". My hasty reaction was to create my temporary data set with a primary allocation of 0 and a generous secondary. And I encountered further unexpected behavior when I learned of VIO and changed my JCL to use VIO for my temporary data set. I exhausted page space. Is EOF now written for new VIO data sets? -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html