Re: [netrexx-pipelines] Migrate OS/2 pipelines to NetRexx Pipes

[<hp...@we...>]

It's WE of Pentecost, time for something more relevant than
variations of Pipelines. But I'll try to reply most of my
questions by myself, might result in some more insight about all
the obstacles of this drudgery.

Am 03.06.2022 um 18:13 schrieb ((ich)):
> [...]
> Now, how does our project go on? You "delivered" already something
> for the first pipeline
>
>> call RxPipe '< "' || fn || '"!nlocate 1 /*/!varset'
>
> but I was not yet able to grasp your solution in detail. So my
> questions are:
>
> i) How to start a NetRexx Pipelines from ooRexx?

In C:\prgm\NetRexx\bin I had renamed the PIPE.BAT to HIPE.BAT what
should allow coexistence of OS/2 Pipelines with NetRexx Pipelines.
Works from the command line of a cmd window.
(I renamed it again to RIPE.BAT, just as objective to bring it to
a maturity worth to publish.)

> ii) how is the content of 'fn' transferred from "here" to "there"?

What's possible from the command line will also work from Rexx,
ooRexx, and so on. But the arguments to RIPE (PIPE.BAT renamed)
must be clasped/surrounded by double quotes. A simplistic test:

> /* ooRexx */
> fn = vilma.ini
> ripe '"<' fn '|term"'

Works. (Quite funny, when I tried this first it failed, allegedly
function 'term' not found.)
When I look into RIPE.BAT there is no procedure, just one single
line to involve Java (to my comprehension). So it should be
possible to set variable ripe to avoid the need of that renamed
batch file.

> /* ooRexx */
> nrh = value("NETREXX_HOME", , "ENVIRONMENT")
> ripe = '@java -cp "'nrh'\lib\NetRexxF.jar;%CLASSPATH%"',
>  'org.netrexx.njpipes.pipes.runner'
> fn = vilma.ini
> ripe '"<' fn '|term"'
Works. So the BAT file is obsolete for this project.

> iii) how must a varset or varload stage look like with (almost)
> all options as the role model?

First I wanted to see if the ooRexx function 'value' works the
same under NetRexx. It will -- hopefully -- be documented in
"NetRexx Language Reference", ISBN 978-90-819090-1-3, which shows
on the title page Mike Cowlishaw and REXXLA as author (I doubt. In
addition, Adobe Acrobat Reader shows as first bookmark 'List of
Tables' which is linked to 'Content' on page upper-case II, there
'List of Tables' is indicated to be on page upper-case VII, a
mouse-klick on this entry does just nothing, while the 'List of
Tables' is in fact on page upper-case VIII. Sure, this could be a
negligible detail, if not details stand for the whole. The
responsible publisher of this edition is mentioned on page
upper-case I, but who involved them? Mike Cowlishaw?)

Back to the subject. I expected a language reference documentation
to be structured according the language, something like: A-general
concept, B-keyword instructions, C-functions, etc. Since I still
have REXX in mind as I know it since VM/SP my assumption was a
fail. Similar to the "clash of the cultures" I encountered with
ooRexx I see it now with NetRexx. Instead of

> say length("ABC")

I have to put it this way:

> say "ABC".length

What a progress! Re-use of old-style REXX routines impossible.

OK, back to the subject.
Additional question: if I would program a VARLOAD stage or alike,
and set there some variables according the data trickling through
the pipe, would those be set within the VARLOAD routine only or
for the NetRexx program which called the pipe in question? Would
this also be possible from ooRexx? Probably not, but is there a
chance to access a "common pool" from both "dialects"? Something
adequate the GLOBALV concept of VM/CMS? Well, that would require
some kind of "pull" in ooRexx after the NetRexx pipe finished, a
"push" of the results (variables set directly) would be more
appropriate.

Sorry, with my nonexisting knowledge of NetRexx I'm stuck. At
least I was able to answer two questions on my own. Now it's the
turn for real experts.

Best,
M.