On Sep 5, 2011, at 21:48, Glenn Knickerbocker wrote: > On Mon, 5 Sep 2011 12:46:35 -0600, gil wrote: > >> o When I use "pipcmd" and feed it an "addpipe", the subroutine >> pipeline runs asynchronously. I can do other things while it >> runs. But it only works once. If I try it again, I get: >> >> FPLSSU139E No connection available to redefine for * > > Right, that's what all the extra stuff inside the SPECS in my earlier > example was about. Each ADDPIPE has to define a new connector for > PIPCMD's output so the next one can connect to it. This is the same as > you'd do if you were adding data to your own stage's output: > > addf = 'addpipe (end ?) *.out: | f: faninany | *:' > adddelay = addf '? literal +2 | delay | f:' > addquery = addf '? query | f:' > > adddelay > addquery > 'output something else' > > Feed the same stuff to output X in your example for PIPCMD to execute: > > 'callpipe var adddelay | *.out.X:' > 'callpipe var addquery | *.out.X:' > I tried something like that, but outside the pipeline. Then I ran it in a loop to test it. It failed with resources exhausted after a few thousand iterations.
I didn't know it was possible to refer to labels once a pipeline had been created. I'll experiment some more. Thanks, gil
