Wait...instead of focusing on details (like the use of %RCR), why not ask a more basic question: What can be done to optimize Taskman. Bhaskar has raised this issue a number of times, saying that the current architecture is far from optimal on GT.M (and probably Caché, too). I agree that just using JOB would be more efficient than the current approach of maintaining a pool of submanagers. But how is it possible to maintain the currwent interface. Well, one possibility would be to have a routine that would serve as the Taskman *prelude*, something like:
PRELUDE ; code to initialize the environment and set a Taskman specific error trap (so errors can be logged and handled in Taskman) D @routine cleanup code goes here Q So far, so good, but what about controlling the number of running tasks? A natural approach would be to use a counting semaphore, initialized to the task limit. Each new task would execute a DOWN on entry, and UP upon exit (normal or abnormal). As you know, DOWN on a semaphore with a count of 0 blocks until another process executes an UP. I don't know what support is built into GT.M, but semaphores are certainly supported by Linux (or any modern Unix). === Gregory Woodhouse <[EMAIL PROTECTED]> "Judge a man by his questions not his answers." --Voltaire ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Hardhats-members mailing list Hardhats-members@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hardhats-members
