On 2014-06-07, at 11:17, retired mainframer wrote: > Bob quoted the JCL manual. So yes, DUMMY is usable with QSAM. > > The issue however is what is IEBCOPY's requirement. My manual says SYSUT2 > must reside on a device that supports QSAM, such as (but not limited to) > DASD or tape. Since a DUMMY dataset does not reside on a device, it fails > that test. > Which manual and section? This depends on some nice semantics of "reside on a device". Why, for example, does it permit VIO and SYSOUT but prohibit DUMMY and UNIX files? The latter, at least have to "reside" somewhere. And it's arguable whether VIO files "reside on a device".
> The fact that the error message refers only to DASD and tape is just another > example of poor wording. > Indeed. > Why IEBCOPY has such an apparently unneeded restriction is a different > question. > Indeed. I think of Doug Gwyn's maxim (out of context here): "Unix was not designed to stop its users from doing stupid things, as that would also stop them from doing clever things." Sometimes it seems to me that z/OS more effectively stops users' doing clever things than their doing stupid things. > And the original issue about member lists is solved by using VIO instead of > DUMMY. > Which has lower overhead, VIO or SYSOUT? -- gil
