Try CALLPIPE (propagate) to get the EOF to flow out of the primaary pipe to the secondary.
On 16 January 2013 04:21, Glenn Knickerbocker <[email protected]> wrote: > I was surprised to get a stall at EOF from this simple CALLPIPE: > > 'callpipe (end /) *: | sql nocommit execute | *:' , > '/ *..1: | sql nocommit execute | *..1:' > > when output 0 was fed back into input 1. The stall doesn't have > anything to do with SQL; it happens with any stages. If I use ADDPIPE > instead, it terminates fine. With CALLPIPE, I have to include something > like GATE to sever the fed back input at EOF. It can be in the calling > pipeline: > > 'PIPE (end /)' input '| 0: chop 0 | take last 1 | stop: gate' , > '/ 0: | a:' mystage etc. '| a: |' output > > or it can be on the input side of the CALLPIPE: > > 'callpipe (end /) *: | 0: chop 0 | take last 1 | stop: gate' , > '/ 0: | insert ,0, | *:' , > '/ *..1: | stop: | insert ,1, | *..1:' > > It doesn't help if it's on the output side, though. This still stalls: > > 'callpipe (end /) *: | 0: chop 0 | take last 1 | stop: gate' , > '/ 0: | insert ,0, | stop: | *:' , > '/ *..1: | insert ,1, | *..1:' > > ¬R
