From: Gang Zhao (galaxy_at_cs.washington.edu)
Date: Wed Jan 07 2004 - 15:59:07 PST
This paper describes a system for shared use of a computer system.
The author, Dijkstra takes a much more informal tone in this paper than he does in his books.
His offhand comments on the sharing of half-time people are reminiscent of analysis done in the book "The Mythical Man Month".
The claims about verifiability, the apparent lack of a need for testing, and the comments about provability of correctness are interesting. If the paper were written by anyone other than Dijkstra, I would have dismissed these comments and further claims in the paper. But Dijkstra's book "A Discipline of Programming" gives him credibility on these topics.
The section "A Survey of the System Structure" appears to be exciting. This is an early reference to the joys of virtual memory. Given the limited (32k) memory, this can be seen as a fundamental breakthrough.
Like many computer science papers, this one makes the assumptuion "'delaying the progress of a process temporarily' can never be harmful to the interior logic of the process delayed", which makes the assumption that the process has no notion of time, does not comminicate with I/O or other processes having a notion of time, and does not interact with the user.
The system hierarchy discussed gives good reasons for the various levels and the layers of abstraction that they imply. Particularly important is the fact that only the level 0 process must deal with interupts, and only the level 0 and 1 processes must deal with the details of virtual memory.
The terminology used in the section "Design Experience" including 'short shots' is revealing about the time that this was written, during the main thrust of the Appolo rocket missions.
I was unable to fully understand the proofs of Harmonious Cooperation. It appears that details have been omitted.
This archive was generated by hypermail 2.1.6 : Wed Jan 07 2004 - 15:59:12 PST