While this sounds intriguing, neither "IF" nor "PIPCACHE" are documented
in the z/VM 5.2 CMS Pipelines Reference (SC24-6076-02). Are they stages you wrote? Brian Nielsen On Thu, 13 Jul 2006 21:51:40 +0200, Rob van der Heij <[EMAIL PROTECTED]> wrote: >My apologies to the non-plumbers - we should have this kind of threads >on CMSPIP-L instead... > >On 7/13/06, Brian Nielsen <[EMAIL PROTECTED]> wrote: > >> Not my goal, but thanks anyway. I was hoping to eliminate the trivial > >Something that may help in a few cases (which I only dare to suggest >now that Melinda is travelling) is the IF stage. It does not help with >stages that split the entire stream (like take, frlabel, etc) but it >does help when some of the records need to bypass a pipeline segment. >Like this: > >| x: if pick w1 == ,USER, >| xlate w3 00-FF * >| x: > >Normally this would require an extra faninany and the dangling >connection stuff at the end. > >And sometimes it is handy to negate the selection use the NOT stage so >that you can code the secondary sooner and pick up the main stream >without the risk of forgetting the remainder. Especially when you put >temporary debugging stuff in, it helps to limit the number of places >that you need to modify to take it out again. > >| ... >| c: not count lines | spec ,We have, 1 w1 nw ,records, nw | cons \ c: >| ... > >Or something like this: >| .. >| f: fanout | insert ,Trace #1 , | cons \ f: >| .. > >And when you code your own multi-stream stages in REXX, you can take >advantage of an unconventional way to arrange the the streams. >My favorite one is the generalized pipeline cache that fits around >some cpu intensive plumbing and memorizes what goes in and out that >segment, and bypasses it when the item is in the cache. > >| .. >| p: pipcache >| ... expensive plumbing >| p: >| > >Rob >-- >Rob van der Heij >Velocity Software, Inc >http://velocitysoftware.com/ >======================== ========================= ========================
