TheHouston Automatic Spooling Priority Program, commonly known asHASP, is an extension of theOS/360 andOS/VS2 (SVS) operating systems, providing extended support for "job management, data management, task management, andremote job entry."[1] It was the basis forJES2 inMVS.
OS/360 includedspooling routines, calledreader/interpreters andoutput writers.[2] Each reader/interpreter was "responsible for reading one input job stream" – that is one input device. Likewise each output writer was responsible for controlling one printer orpunch. Spooled data were stored in OStemporary datasets controlled by standard OS services. Each reader/interpreter or output writer was a separate operating system task in its own partition or region.[a] A system with a large number of readers, printers, and punches might have a large number of spooling tasks.
HASP was developed byIBM Federal Systems Division contractors at theLyndon B. Johnson Space Center inHouston. The developers were Tom Simpson and Bob Crabtree. HASP was a program that ran on a mainframe, and performed functions such as:scheduling, control of job flow,spooling and printing/punching. HASP had no support forIBM System/360 Operating System Remote Job Entry, 360S-RC-536, but provided roughly equivalent facilities of its own.
In HASP II V3, Roger Fajman, Bill Jones, and Jim Oberthaler ofNIH created the shared spool capability for HASP that was used by many mainframe sites. It allowed each HASP system to share a common spool and checkpoint. This enabled workload balancing in a multi-mainframe environment. In HASP II V4, Don Greb and Dave Miko ofMellon Bank moved HASP shared spool to this version. Don Greb of Mellon Bank carried it forward intoJES2 multi-access spool (IBM's formal support of HASP inMVS). Over 350 copies of the HASP II V4 shared spool mods were distributed around the world by Mellon Bank. The shared spool Mellon Mods were added to the SHARE distribution process so they could be more widely accessed. The "Mellon Mods" were the base for Jes/2 Multi Access Spool.
The program was sometimes referred to under various other names, but there is no indication of IBM ever using them in official documents.
The program became classified as part of theIBM Type-III Library. It had a competitor,ASP (Attached Support Processor), which ran on one mainframe and controlled scheduling of other attached mainframes. ASP later becameJES3.
InMVS, HASP became JES2, one of twoJob Entry Subsystems. It was many years before the HASP labels were removed from the JES2 source, and the messages issued by JES2 are still prefixed with "$HASP".
A modified version of HASP was used to provide batch spooling and remote job entry services for theMichigan Terminal System during the late 1960s, 1970s, and 1980s.[3][4]
HASP bypassed most operating system services with code specially tailored for efficiency. HASP operated as a single operating systemtask[5]and usedcooperative multitasking internally to runprocessors to perform tasks such as running card readers, printers, and punches, managing thespool files, communicating with the system operator, and driving multiple communication lines for remote job entry.
Hasp was written entirely in System/360 assembler and atypical HASP system might require 86KB of memory, which could be virtual memory on OS/VS systems.[6]
![[icon]](/mt/?noimg=&dark=on&url=https%3a%2f%2fen.wikipedia.org%2fwiki%2fFile%3aWiki_letter_w_cropped.svg) | This sectionneeds expansion. You can help byadding missing information.(November 2025) |
HASP uses one or moreSPOOL volumes to contain jobs under its control. HASP stores spooled datasets intrack groups on the SPOOL volumes, rather than as datasets known to the OS. ASysgen for OS/360 or SVS must specify a pool of pseudo-devices,[b][7]typically on different channels from the real devices. The access methods build the same channel programs as they would for real card readers, card punches and printers, but HASP intercepts theEXCP requests from the access methods. As HASP is reading the input stream, it modifies and stores the JCL on SPOOL volumes and stores eachSYSIN data set separately. When HASP is ready to run the job, it passes the job to theReader/Interpreter and an availableInitiator] selects it.
HASP recognizes each explicit DD *, explicit DD DATA and implicit DD * dataset and stores it on SPOOL volumes. During execution of a step, HASP assignes each SYSIN dataset to a pseudo-reader that the access methods treat as a real card reader, and intercepts the EXCP requests from the access method in order to provide card images from SPOOL. HASP simulates aunit exception after providing the last card image in the dataset.
HASP changes eachDD SYSOUT= data set as a reference to a pseudo printer or pseudo card punch. HASP intercpts each EXCP from the access methods, storing each card image and line image on SPOOL. After the job completes, a HASP printer, punch or remote processor that matches the selection criteria of a SYSOUT dataset copies it to an output device.
Much of what would becomeJES2'sJECL was a part of HASP.[8]HASP overloadsdelimiter statements (/*) as HASP control statements.
Hasp originally supported IBM Synchronous Transmit-Receive (STR) batchterminals; however, HASP II dropped support for STR and only supported theBinary Synchronous Communications (BSC) protocol. HASP II supportedIBM terminals such as2770,2780,3770,3780,but additionally provided support formulti-leaving[9]communication with intelligent workstations such as theIBM 1130,IBM System/3 and theSystem/360 Model 20. Multi-leaving is "fully synchronous, pseudo-simultaneous, bi-directional transmission of a variable number of data streams between two or more computers utilizing binary-synchronous communications facilities."[10]
Third party vendors developed a variety of 2770/2780/3780 simulators and Multi-leaving implementations for use with HASP and ASP. Some of these vendors incorrectly referred to their products as HASPemulators, but the products were actually terminals that talked to HASP; they did not perform any of the functions of HASP.
The RJE console operator can enter a restricted set of HASP Console Operator Commands, each beginning with a $ (dollar sign).
Operands were restricted to those from this remote site, hence DJ1-999 (below) would only display typically a few, those still outstanding from the operator's remote site.
Some of the HASP RJE Console commands, which could be entered in short (single letter) form, or spelled out, are:
| Short | Long | Operand(s) | Example |
|---|---|---|---|
| E | restart | device_name | E rm1.rdr |
| Z | stop | device_name | Z rm1.pr1 |
| D | display | (J,I,& others) | e.g. |
| DJ | display job | Job Number (or range) | DJ345 or DJ1-999 |
| DA | display active | (displayed active jobs) | DA |
| DI | display initiators | (displayed initiator letters) | DI |
| C | cancel | (Job or device_name) | e.g. |
| CJ | cancel job | Job Number | CJ678 |
| C | cancel | device_name | C rm1.rdr |
A command line "C rm1.rdr" was meant to cancel rather than submit the current deck of cards being read. This might have been of use if a card jam made it more sensible to let other jobs go ahead while one or more cards were replaced at a nearby keypunch.
HASP job log output provided a summary of the resources used for the job (output appeared in all caps):