On Aug 31, 2011, at 13:35, Glenn Knickerbocker wrote:
> On 08/30/2011 09:23 PM, Paul Gilmartin wrote:
>> 'addpipe *.input.C: | faninany | *.input.B:'
>
> There's nothing to connect to on the other side of input C. A dangling
> pipeline segment like this has to go from output to input so that it's
> connected to your stage at both ends:
>
> 'addpipe *.output.C: | faninany | *.input.B:'
>
> That's not going to give you the result you're looking for when you
> ...
> So if your goal is to avoid having to write a new CALLPIPE for each
> combination of dynamically added input streams, this won't get you
> there. Once you've connected them downstream with ADDPIPE or SHORT,
> they're not available for you to sever and reconnect later.
>
Hmmm. Sounds as if there's no clean way to do it. I've been
connecting my asynchronous stage to:
specs /MESSAGE * / 1 1-* next | cp
then fielding the output with STARMSG. That's not clean; it's dirty.
Perhaps if I:
'addpipe *.output.C: | runpipe | faninany | *.input.B:'
(needs embellishment). Then any stage description I write
select output C
output some-stage
should run asynchronously, and I can capture its output from B:.
Need to try it tomorrow.
Thanks,
gil
