On 2019-03-23, at 05:22:22, Rob van der Heij wrote:
>
> I think the concern is that people are using the name "CMS/TSO Pipelines"
> for something different. This is confusing; we don't like people doing
> that. If you write something that implements a data flow programming model,
> nobody will hold you back and some may encourage you. Just don't claim it
> is CMS Pipelines.
>
Is "Hartmann Pipelines" preferable?
> ..., I find piping the the UNIX shell just
> "a toy" as John says. Some are just fine with what they have. I did look at
> njpipes some time ago and within minutes I was in trouble (meaning a pipe
> that doesn't do what I'm used to with CMS Pipelines).
>
OTOH, I find CMS Pipelines syntax painfully cumbersome. Instead of:
{ stage1; stage2; stage3; } | stage4 # I must code:
CALLPIPE stage1 | APPEND stage2 | APPEND stage3 | stage4
... and I understand APPEND is deprecated for poor performance and I should,
rather, do something with FANIN and connectors for which I need to RTFM.
Every time. Ironically, the paucity of delimiters in Pipelines leads to
comlexity in practice.
I'll grant the argument about concurrent pipelines before anyone needs to
advance it. I've done it, once, in POSIX shell; that was painful. Slightly
better in z/OS Rexx.
> When you write an editor for the PC, you don't call it "TSO" just because
> that's the only thing you've done on MVS and believe that's all that others
> do there.
>
There is a commercial analogue of XEDIT for UNIX:
http://www.wrkgrp.com/
... the command to invoke it needs to be "xe"; "xedit" was already taken by
a competing product.
-- gil
