Incompatible Timesharing System
|
|
it:Incompatible Timesharing System
ITS, the Incompatible Timesharing System, was an early, revolutionary, and influential MIT time-sharing operating system; it was developed principally by the Artificial Intelligence Laboratory at MIT, with some help from Project MAC.
ITS development was initiated in the late 1960s by those (notably the majority of the AI Lab at that point in time) who disagreed with the direction taken by Project MAC's Multics project (which had started in the mid 1960s), particularly such decisions as the inclusion of powerful system security. The name was a hack on the earliest MIT time-sharing operating system, the Compatible Time-Sharing System, which dated from the early 1960s.
ITS was initially developed for the Digital Equipment Corporation PDP-6 computer, and later moved to the PDP-10 once it became available, where it saw the majority of its development and use.
Among numerous interesting features and oddities, the ITS top-level command interpreter was the PDP-10 machine language debugger (DDT), whose commands looked like line noise to the uninitiated.
Its main editor for many years, TECO, was programmable in a similar-looking gibberish. The EMACS editor is a descendant of a collection of TECO macros, now much developed.
Among other significant and influential software subsystems which were developed on ITS, the Macsyma symbolic algebra system is probably the most important; the GNU INFO help system used in Linux, some versions of UNIX, and EMACS was also started on ITS.
| Contents |
Significant technical features
ITS was an operating system with many revolutionary features; for some significant ones, it was the first to deploy them. Among them were:
- It had the first device-independent graphics terminal output; programs generated generic commands to control screen content, which the system automatically translated into the appropriate character sequences for the particular type of terminal operated by the user.
- A general and powerful mechanism for implementing virtual devices in software which ran in user processes (which were called "jobs" in ITS).
- Using this mechanism, it provided transparent inter-machine filesystem access (almost certainly the first operating system to do so). The ITS machines were all connected to the ARPAnet, and a user on one could perform any operation on a file on one of the other machines that they could on files on the local machine.
- Sophisticated process management; user processes were organized in a tree hierarchy, and a superior process could control a large number of inferior processes. Any inferior process could be frozen at any point in its operation, and any of its state (including contents of the registers) examined; the process could then be restarted without any problems. (This facility is now in one of the GNU tools.)
- An advanced software interrupt facility that allowed user processes to operate asynchronously, using complex interrupt handling mechanisms.
- PCLSRing, a powerful and subtle mechanism which provided what appeared (to user processes) to be non-blocking, safely interruptible system calls. No process could ever observe any process (including itself) in the middle of executing any system call.
Many of these, and numerous other significant advances, were later picked up by other operating systems.
User environment
The environment seen by ITS users was philosophically significantly different from that provided by most operating systems at the time.
- Initially there were no passwords, and a user could work on ITS without even logging on. Logging on was considered polite, though, so people knew when you were connected.
- All users could bring the system down, but a message was broadcast to say who was doing it.
- All files were editable by all users.
- All users could talk with instant messaging on another's terminal, or they could use a command (SHOUT) to ask all active users for help.
- Users could see what was happening on another's terminal. The user being watched was informed, and could kill the viewer's session. This facility was later disabled in an interesting way: it looked like the session was killed, but was not.
- Tourists - guest users either at MIT AI Lab terminals, or over the ARPAnet - were permitted. A policy (http://www.art.net/Studios/Hackers/Hopkins/Don/text/tourist-policy.html) was later published on tourist access.
Original authors
Further reading
- Donald E. Eastlake, ITS Reference Manual, Version 1.5 (ftp://publications.ai.mit.edu/ai-publications/pdf/AIM-161A.pdf), (MIT AI Laboratory, 1969) documents a very early version of the system (Warning, very large file, scanned page images)
- Donald E. Eastlake, ITS Status Report (ftp://publications.ai.mit.edu/ai-publications/pdf/AIM-238.pdf) (MIT AI Laboratory, 1972) documents a somewhat later, but still fairly early, version (Warning, very large file, scanned page images)
- Alan Bawden, PCLSRing: Keeping Process State Modular (http://fare.tunes.org/tmp/emergent/pclsr.htm)
External links
- http://www.its.os.org/ contains file system images of various ITS machines, including source and documentation of the final system
- http://www.cosmic.com/u/mirian/its/ provides instructions allowing ITS to run on a PDP-10 simulator.
- http://www.art.net/Studios/Hackers/Hopkins/Don/text/tourist-policy.html (Historical) Tourist access policy in ITS at MIT.
- http://zane.brouhaha.com/~healyzh/_info_/its.primer An Introduction to ITS for the MACSYMA User
- http://zane.brouhaha.com/~healyzh/sysdoc/sysdoc.html ITS System Documentation