> I still don't doubt John G's observation; I merely suspect that whatever utility he tested with detected that some data set was allocated to DUMMY and bypassed processing it entirely.
The assertion was NOT "some programs ignore DCB parameters when DUMMY is coded." That's probably true. The assertion was "JCL ignores DCB= when DUMMY is coded." That assertion is quite apparently false, as your experiment shows. I don't doubt his observation but the conclusion is false. Further, there was an assertion that DUMMY is somehow different from DSN=NULLFILE. I believe that assertion is also mistaken. We have seen no evidence that it is correct, and several people have cited JCL manuals that seem to show that it is not correct. I'm not interested in beating John up (in spite of his gratuitous insulting comment about my post) but I think the record should be set straight for z/OS practitioners who may be trying to learn from this forum, such as the original questioner. And speaking of which, back to the original question: shouldn't BLKSIZE=0 "work" for DUMMY allocations? On the one hand, if you fell to earth this morning and were learning z/OS for the first time, you would ask "what the heck difference does it make? What is the meaning of the block size of a dataset that doesn't exist?" But of course we veterans know that the output doesn't exist but the DCB and the processing logic do, so BLKSIZE matters. (OS/360 should have abstracted and hidden all of this the way UNIX does, but that's a subject for a different thread.) On the other hand, I suspect the IBMers who wrote SDB viewed it as a "match the block size to the device" problem and DUMMY was not on their list of devices. On the third hand, why shouldn't it be? Why shouldn't SDB support a "device type" of DUMMY? Why shouldn't what the original questioner's Sysprog did work? No, I don't want to PMR it, but I would argue logically that it IS APARable. Charles -----Original Message----- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of Paul Gilmartin Sent: Saturday, May 28, 2005 11:02 AM To: [email protected] Subject: Re: DUMMY and BLKSIZE In a recent note, Charles Mills said: > Date: Fri, 27 May 2005 16:16:26 -0700 > > I don't consider the problem interesting enough to try the experiment, > but I think you are mistaken. I am relatively certain that > I would not so impugn either John G's veracity or his observational acuity as to dismiss off-the-cuff his report of an experimental result. > - DCB= is *not* ignored when coded with DUMMY. Witness the original > question. There was a functional difference between DUMMY,BLKSIZE=129 > and DUMMY,BLKSIZE=0. > So, I tried my own experiment. On z/OS 1.5, the job: 3 //STEP1 EXEC PGM=IEBGENER 4 //SYSUT2 DD SYSOUT=(,) 5 //SYSUT1 DD * 6 //SYSPRINT DD SYSOUT=(,) 7 //SYSIN DD DUMMY,DCB=(RECFM=FB,LRECL=70,BLKSIZE=999) //* Produced the messages: IEF236I ALLOC. FOR JOBCARD STEP1 IEF237I DMY ALLOCATED TO SYSIN IEF142I JOBCARD STEP1 - STEP WAS EXECUTED - COND CODE 0012 And SYSPRINT: 1DATA SET UTILITY - GENERATE PAGE 0001 -IEB319I INVALID SYSPRINT/SYSIN BLOCKSIZE So, I conclude DCB parameters are effective on DD DUMMY. I still don't doubt John G's observation; I merely suspect that whatever utility he tested with detected that some data set was allocated to DUMMY and bypassed processing it entirely. ---------------------------------------------------------------------- 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
