On Wed, 6 Nov 2013 12:03:58 -0500, Tony Harminc wrote:
>
>>> IKJ56709I INVALID DATA SET NAME, /etc/logs
>>>
>> Actually, that could be a valid data set name with
>> DISABLE(DSNCHECK) in effect. It appears to me that SMCOPY
>> hasn't been updated to support the new facilities of catalogs.
>
>... Allowing just a single
>name space for classic dsnames and UNIX files is a mixed bag. In some
>cases it's very handy. In others, deciding on syntax alone can make
>for real ambiguity.
>
I largely agree; even as JCL wisely distinguishes between PATH=
and DSN=. I would make an addition of a virtual filesystem,
perhaps such as:
/dev/Legacy/SYS1.MACLIB(SPLEVEL)
... to allow UNIX utilities to manipulate Legacy data sets with
constructs such as redirection, now not supported. E.g.
cat <'/dev/Legacy/SYS1.MACLIB(SPLEVEL)'
Support of inode entries such as size and permission bits remains
problematic, even as it is for legacy data sets served by NFS.
The "//" hack has long been deprecated by POSIX. IBM should
have heeded that caution and avoided it.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
