On Fri, 29 May 2009 20:02:54 +0000, Ted MacNEIL wrote: >>And since then, and yet, I wonder why IBM doesn't DTRT and modify allocation >>to unconditionally write an EOF, regardless >of DSORG, SMS, whatever. Are they fearful of the performance impact? > >What's the EOF on a PDS? >PDSE? >Indexed VSAM? > It doesn't matter. It's always possible that the newly allocated extent contains at its beginning an EOF, residual data from previous use. In each of the three cases you cite, the access method must be able to deal with a new extent beginning with an EOF. The techniques that deals with a residual EOF should equally successfully deal with an EOF written immediately previously by allocation.
>There are issues! > Minor ones: o I already mentioned the cost of running a channel program to write an EOF, possibly unnecessarily. o Of course if the primary extent has size 0, it's best not to write the EOF. o Data recovery scenarios. It might be better not to write the EOF if the job specifies absolute track addresses. -- 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
