On Fri, 1 Apr 2011 13:55:49 +0200, Hunkeler Peter (KIUP 4) wrote: >>Afaik delete is done at unallocation of the dd. So the free command >>should follow the allocate one. But why not just deleteing the >>dataset without first allocating it? > My expectation was that if the DSN preexisted and was catalogued allocation would create it uncatalogued (I didn't specify CATALOG) on a different volume. My HLQ is not SMS managed. z/OS 1.12.
My objective was to synthesize atomic behavior (oxymoron?) by holding an ENQ EXC over the interval comprising the DELETE and the RENAME so no other job could perceive an interval during which the data set did not exist. The redundant IEFBR14 COND=(0,LE) step had the same intent. >I was trying to recreate the situation gil describe in his original >post. My first thought was the delete will not work because the >data set is allocated at this time. However, when running the job, I >was astonished to see the delete working. > I was astonished that all references were to the same volser, and that in each case the disposition was KEPT. >As seen here the job doesn't make sense anyway; it was part of >a simplified setup gil made to make some test. And his astonishment >was that the job *was* looping. > Indeed. But even nonsensical constructs should produce predictable results. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
