This paper describes Cedar, a Xerox PARC research system. This ambitious project, begun in 1978, created a single-user system for experimental programming. It was derived from Mesa, but has also borrowed ideas from Smalltalk and Interlisp-D. Like them, Cedar creates a network-based workstation, with a high-resolution display and a mouse. By 1983, Cedar offered: a powerful language; packages ranging from graphics to database management; a breathtaking, typeset-quality editor (Tioga); a distributed file system; source-level debugging; and many other nontrivial facilities. It manifests the following: concurrency within and among applications; industrial strength facilities, suitable for large, complex systems; and integration with respect both to components and to usage paradigms.
Cedar reduces many good ideas to practice, including such philosophical dicta as The Law of Least Astonishment (user experience should correctly predict system behavior), Do What I Mean (interaction mistakes, e.g., spelling errors, should be automatically corrected in context), and Unlimited Undo (every action should be reversible).
The paper under review simulates a live Cedar demonstration; it is, in fact, based on a videotape. The reader is led through realistic programming tasks--browsing documentation, finding and fixing bugs, reading electronic mail, designing a new icon--with 27 screen snapshots and commentary. This paper, condensed from an internal PARC report, also appeared in [1] and [2].
This paper reflects PARC’s usual high standards: it is clear, attractive, convincing. It makes us wish that Cedar were generally available] Of course, PARC’s insularity colors it, as it colors many PARC efforts. Modern scientific inquiry presumes repeatable experiments and verifiable results. Cedar seems ideal; but can we test its claims? Not easily; Cedar runs on a proprietary workstation (as fast as a large VAX), so we must accept what is reported. Yet this work shows that new ground has, again, been broken.
This lively, entertaining paper shows that Cedar is not only practical, but fun to use. It has a few minor flaws:
(1) Gaps were left in its abridgement from a longer version.
(2) The figures--screen images--are cluttered with windows of text, menus, and icons; they show Cedar’s high-bandwidth dialogue, but the point being discussed is sometimes lost. This is especially true in [1]; its figures have been over-reduced, making the details unreadable.
(3) Some Cedar examples in the text, unsupported by figures, are obscure.
(4) There is not enough detail about the Cedar language. This last point is important: although crucial to an understanding of the environment, language details are only found in internal PARC reports.
Few hypotheses are offered; this is primarily a commentary on a past research effort. It documents, explains, and justifies the research. As such, it is effective. In another and more important sense, it supports Cedar’s key implicit hypotheses: That a better development context yields better systems; and that an integrated environment yields cooperating applications. Everyone should at least scan this paper, to see what a programming environment can be. It gives a good overview of Cedar. Cedar is well worth a look. Sadly, this is as close as most of us will get.
