On 09/07/2011 01:38 PM, Paul Gilmartin wrote: > But why the PIPWAR1365E?
Hmmm, I ran into that but only intermittently, and must have assumed I'd just made some typos. It looks like you may actually have to SUSPEND the current stage to force WARP to run and register the warp before you try to connect anything more to it. I don't get why COMMIT 0 isn't sufficient--even when WARP is connected to another stage in the current pipeline set, or even the current stage: 'addpipe warp cons | cons' /* disjoint, might expect trouble here */ 'addpipe warp out | *:' /* connect to next stage downstream */ 'addpipe warp in | *.in:' /* connect to *this* stage */ 'nocommit' /* do next CALLPIPE without committing */ 'callpipe warplist | cons' /* doesn't list any warps */ 'commit 0' /* this should wait for everyone else */ 'callpipe warplist | cons' /* Surprise! still no warps */ 'suspend' /* wait for everyone else to run */ 'callpipe warplist | cons' /* now we get the whole list */ ¬R
