On Tue, Dec 9, 2008 at 8:19 PM, Glenn Knickerbocker <[EMAIL PROTECTED]> wrote:
> So the pitcher really doesn't ever delay the record at all, but the > catcher may delay it indefinitely, right? I'm not sure what you mean. The idea is that a pipeline segment like this .. | warp test \ warp test | .. does not delay the record. One of the practical uses of the pipeline warp is for example a centralized logging stage like this: \ warp syslog | time | > logging file a Now when you spawn pipelines with addpipe for example, they can simply use "..| warp syslog" to deposit their logging in the central service. That avoids the need to generate a cascade of faninany stages to catch it. I have also done some nifty things with instore / outstore where you throw the token through the pipeline warp and let the catcher unpack the stored stream (and if you duplicate the token you can get multiple copies of the data). I was very impressed to see this work out of the box. -Rob
