Thanks for the explanation Rob. -- bc On Sat, Nov 1, 2014 at 4:02 AM, Rob van der Heij <[email protected]> wrote:
> On 1 November 2014 02:28, Bob Cronin <[email protected]> wrote: > > > Thanks, the copy does solve the stall. > > > > > while that is still holding that record on it's secondary output > > > > I guess this is what I don't understand. Why is it still holding that > > record exactly? > > > > It's because "gate" and "totarget" do not delay the record, so when you > divert that terminator record to the primary input of "gate", it is still > being held by "totarget" and the secondary output of "gate", and "gate" > will not turn around to inspect its primary input for a record until it > gets rid of the current one. The "copy" lets you park the record briefly > and frees up the secondary of "gate" again. > > Rob > > PS Spotting that you need a "copy" is one thing - being able to predict > exactly where it goes is more of a challenge. I also have a pipe that needs > two consecutive "copy" stages in one segment (that kills the method of just > trying where it needs to be). And "elastic" often also solves the stall, > but comes with a cost (leaky pipes). >
