It was for this modularization purpose that the EXEC and PCP subsystems were created. To the extent possible, these subsystems are meant to orchestrate the other subsystems while shielding them from arcane considerations that arise in the design of distributed processing systems. Early investigations into methods for implementing the EXEC and PCP subsystems led (among other areas) to consideration of "perl", a scripting language with extensions that might facilitate subsystem communication over a multiple-machine configuration. The study of perl was undertaken by T. Conrow and J. White. The study involved familiarization and exploration.
After several weeks of study, the rate of progress in mastering perl left room for concern about whether this approach would bear fruit on the time scale needed, and so some priority was given to discussing the work done so far, future plans, and possible alternatives. T. Conrow reported on the particular aspects of perl that could be useful for 2MAPPS and described experiments currently underway. The potentially useful aspects of perl involve the distributed-system communication capability (which could enhance the effectiveness of the MONITOR function of EXEC, as well as provide a mechanism for launching tasks on multiple machines) and a convenient way (once set up) for subsystem development to toggle between a subsystem test environment and a system test environment. On the other hand, it was found by both investigators that perl is something of an opaque and alien language, raising fears about its maintainability.
Since J. White (the cognizant engineer of the EXEC and PCP subsystems) is currently scheduled to be working on the TAPELOAD subsystem, and since T. Conrow is available to 2MASS at a significant level for the next year or so (but not available to assume cognizant engineer responsibilities), it was decided that T. Conrow will attempt over the next two or three weeks to develop a perl demonstration script that performs the functions needed by 2MAPPS, and if this can be done, we will continue to base the EXEC and PCP design (primarily PCP) on perl. The design and implementation would be done primarily by T. Conrow and delivered to J. White, accompanied by a thorough tutorial. The demonstration script would launch and communicate with at least two dummy subsystems running on at least two different machines. The dummy subsystems would need to demonstrate that they had received their command-line parameters successfully. The demonstration activity will not preclude investigating other possible avenues to achieving the functional capabilities needed.