On Tue, 21 Dec 2021 07:41:08 -0600, Bill Schoen wrote:
>Yes, this implies open() and all other file related APIs.
>
tell()? seek()? Support as either binary or text?
Long ago on MVS-OE, you had pondered the possibility to that end.
It appears something similar id coming to pass.
>It will use the ISPF ENQ model as documented in the ISPF Customization book,
>
Excellent!
>however, ENQs may be held for a short period of time after last close of a
>file.
>
Whether on a cosmological or atomic time scale. Implying that close() amd
DEQ can't be fused into an atomic operation. IIRC, I've observed similar
behavior of NFS. I suspect NFS delays DEQ for performance reasons.
I wonder whether a mischievous user can create a deadlock with:
cp a b & cp b a # ? Or:
cat /dev/fd/0 >"'data.set(member)'" # and typing slowly.
Naming convention? Some users will exist that "//'data.set.name'" be
preserved, at least as an alternative.
TSO prefixing? (Which I've always deemed a feeble stopgap for chdir().)
>On Tue, 21 Dec 2021 09:57:54 +0800, David Crayford wrote:
>> ...
<https://www.ibm.com/common/ssi/ShowDoc.wss?docURL=/common/ssi/rep_ca/0/897/ENUS221-260/index.html&lang=en>
>>*New file system support to access z/OS data sets*
>>
>>IBM plans to provide a new file system type that will render traditional
>>z/OS data sets accessible by the z/OS UNIX name space. This will enable
>>z/OS UNIX applications, tools, and utilities transparent access to data
>>in these data sets in a secure and consistent manner.
Thanks,
gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
