On 2018-09-04 21:01, Andrew Rowley wrote:
ISPF also does a SYSDSN ENQ on the dataset.
IEBCOPY, IEBGENER etc. do not do ENQs - you write the JCL to do the ENQ.
If you are updating a dataset you are supposed to use DISP=OLD, which
will protect you against concurrent updates, including ISPF edit. Many
use DISP=SHR because it is more convenient, but then it is your decision
to give up the protection of DISP=OLD - whether it is from ISPF, or
other instances of IEBCOPY, IEBGENER etc.
There are many examples where updates using DISP=SHR can cause problems
including data loss. If the customer chooses to do that they own the
problem.
Is there any IBM software that handles its own ENQs (as opposed to
relying on what the customer codes in JCL) that is not compatible with
ISPF and it's use of SYSDSN and SPFEDIT? My gut feeling is that if there
was it would be APARable, but I don't know.
Absolutely, IEBCOPY, IEBGENER, et al, do not do their own SYSDSN ENQs.
These are provided by the system based on what is coded in JCL or
dynamic allocation, and that is what makes them reliable when used
properly. Like you say, using DISP=SHR when it is not appropriate to do
so is a great way to lose data.
Unlike SYSDSN, SPFEDIT is not taken care of by the system. There are
plenty of IBM and non-IBM programs that do not follow the SPFEDIT ENQ
conventions, and I don't think any product other than ISPF itself would
be APARable for failing to do so. Since it is not universal, it does not
carry the same level of reliability. It's like a gentlemen's agreement
in a world where not all are gentlemen.
--
Regards, Gord Tomlin
Action Software International
(a division of Mazda Computer Corporation)
Tel: (905) 470-7113, Fax: (905) 470-6507
Support: https://actionsoftware.com/support/
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
