On Tue, 10 Jan 2017 07:27:12 -0600, John McKown wrote:
>OK, I know that there can be multiple TIOTs in a single address space. I
>know this because the documentation the IEFUJI says that it can do I/O to a
>data set either by doing a DYNALLOC on it, or by putting the DD in the
>initiator procedure.
>
But don't those go in the same TIOT?
I just looked at an actual control block:
z/OS V2R2 MVS Data Areas Volume 4 (SCF - XTL)
TIOT Information
TIOT mapping
...
01 POINTED TO BY:
TCBTIO field of the TCB data area
... (and 4 others) ...
So the TCB points to the TIOT. That's the first (small) step toward
allowing, mutatis mutandis, different TCBs to identify different TIOTs.
>So, have you ever wanted to run multiple, concurrent, executions of some
>program which does not accept an alternative DD name list? For instance,
>some in-house COBOL program. Wouldn't it be kind of nice to be able to
>specify some parameter, such as NEWTIOT=YES so that the program being
>attached would have its own, "private" TIOT & XTIOT, which would then
>propogate to any of its children TCBs which did not have a NEWTIOT=YES.
>Hum, it might even be nice to be able to list which DDs are to be "passed
>down", ...
>
Passed and possibly renamed, such as passing a DDNAME dynamically
allocated by the parent to be known as INPUT by the child and removed
from the parent's TIOT. (when using SYSCALL spawn, one of my
recurrent errors is failing to close() a descriptor in the parent.)
> ... a bit like you can say whether or not to share a specific SUBPOOL
>with a subtask. That might be nice for something like SYSPRINT.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
