> On 2019-09-12, at 14:26:36, Seymour J Metz wrote:
>
>> This implies that submitter must have an OMVS segment.
>
> No. If he can run in a Unix shell than he can use the Unix functions for
> REXX.
>
> I looked at Chapter 2. OMVS, a 3270 terminal interface to the z/OS shell in
> z/OS: UNIX System Services User's Guide and didn't see anything about
> requiring an OMVS segment. It's possible that you need an OMVS segment to use
> the OMVS command, but that's certainly not implied by the functions for RXX
> under a Unix shell.
>
Usuing any UNIX facility causes the task to be dubbed with the UID in the
owners OMVS segment, a generated unique UID, or (obsolete) a default UID.
>> And it was done as a function package, so SIGNAL ON NOTREADY is not
>> supported.
>
> The ANSI specifications call for returning an empty string rather than
> raising a signal. But I wonder whether IBM would accept an RFE to allow
> external to raise a signal.
>
I find NOTREADY in the Preliminary ANSI Standard at:
http://www.rexxla.org/rexxlang/standards/j18pub.pdf
Have you a better citation?
What should LINEIN() do to distinguish an empty line from end-of-file?
I've used implementations where the STREAM() function makes the distinction.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
