On Fri, 11 Jul 2014 00:32:02 -0400, Robert A. Rosenberg wrote:

>this is also fall-out from the allocation ENQs being
>broken for years. The RNAME being used is DSN NOT DSN+VOLSER like
>ISPF does. There is NO valid reason to issue an exclusive ENQ on a
>dataset name without qualifying it by what volume it lives on. Just
>because I want exclusive access to X.DSN on VOLSER1 does not require
>that access to a X.DSN on VOLSER2 be denied. As I said ISPF has
>solved this issue for EDITs by having its own QNAME and having the
>RNAME include the VOLSER that the dataset you are editing resides on.
>This has been a design flaw that has existed ever since Shared DASD
>and SYSTEMS ENQs have been supported and IBM has done nothing to fix
>the problem by revamping the ENQs done by allocation.

I believe that you are mistaken as to what IPCS does. The IPCS ENQ 
includes member name, not volume serial.

You are also mistaken about the validity of using an ENQ on a data set name 
without volser. The SYSDSN ENQ was defined in the days of OS/360, and it 
was clearly documented for anyone to use.

Because of this, any attempt to change the SYSDSN ENQ to include VOLSER 
would have lead to ENQ not doing what it was supposed to do because of 
other programs that would have had to be changed at the same time as 
the operating system change. That includes, for example ISV programs, 
user programs and CBT programs. Making a change such as you suggest while 
maintaining compatibility would have been a difficult one, not only for IBM, 
but for every customer running the operating system.

It is easy to look back in hindsight and say that the SYSDSN ENQ should 
always have been on DSN and VOLSER and MEMBER name. That was not 
the original design though, and it is not as easy as it sounds at first glance.

One set of difficulties in the using VOLSER in data set enqueues has to do with 
data sets that span multiple volumes. Do you include all volumes in the ENQ? 
What happens if someone mistakenly (or intentionally) requests just the first 
volume of a multiple volume data set? Or just the second volume?

The kind of change that you think that IBM should have made long ago would 
have been practical only if the operating system was the only program that 
could issue an ENQ on SYSDSN, and then only in a single system environment.

-- 
Tom Marchant

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to