From: ···@sef-pmax.slisp.cs.cmu.edu
Subject: The Igor Project
Date:
Message-ID: <CBw0wL.IJG.1@cs.cmu.edu>
Igor is a new ARPA-funded research project in Carnegie Mellon University's
School of Computer Science. I am the project leader, and it is staffed by
the wizards of the former CMU Common Lisp project, plus a few new faces.
We have been working informally on Igor and Dylan since last April, while
finishing up some loose ends on CMU CL. As I announced earlier, CMU CL
will still be available, and we will continue to support it, though at a
considerably reduced level of effort.
Our primary goal is to develop the Igor programming environment, which will
support (and be implemented in) the Dylan language. Igor will also provide
tools for interfacing Dylan code to libraries and routines written in other
languages, but Dylan will be the "universal glue". Igor is designed to
support the ongoing evolution of a software system throughout its life
cycle, from earliest prototyping through deployment and beyond. By
remaining in a single language environment throughout the software life
cycle, we minimize the loss of design information and the need to recover
this knowledge later through costly "re-engineering".
In order to achieve this goal, Igor must provide both an excellent dynamic
development environment and the ability to deliver runtime code modules
that are competitive in size and speed with the code produced by C,
Fortran, or Ada. This is made possible by Dylan's clear separation of the
development and runtime environments -- something not present to the same
degree in Common Lisp.
The Igor development environment will be built around a concept we call
"hypercode" (analogous to "hypertext"). Code in Igor will not be a linear
string of ASCII characters, but rather a complex data structure linking
together routines, class definitions, comments, specifications, diagrams,
test code, edit histories, configuration info, and more. The programmer
will be able to view and browse this hypercode at many levels of detail,
and the code definitions can be presented in whatever order makes the most
sense at the time. (In some ways this is reminiscent of the old Interlisp
environment.) An extensive library of classes and functions will also be
available, with librarian software to guide users in finding what they
needs.
While software can be run and tested within the Igor development
environment, the user can ask at any time to compile the current system
into a "delivery module" that is as small and compact as possible. Once
created, the delivery module will never be changed (though it may be
replaced). All changes are made and tested back in the development
environment. Since the delivery module is not meant to be altered, it can
be welded together as tightly as object code coming from a conventional
static language. At this point, code speed just becomes a question of good
compiler technology. The delivery module can be small, since it contains
only those library modules (or parts of modules) that the system needs at
runtime. It might or might not include a garbage collector.
About half of the effort in this project will go into building the Igor
environment and half into developing a high-quality implementation of
Dylan. In building our Dylan and the associated libraries, we will draw
heavily on code developed for CMU CL, including chunks of the Python
compiler. In addition to our own code, we will establish an online library
for Dylan code that others would like to make freely available to the
community.
Sometime this fall we will release a prototype Dylan implementation
embedded in CMU CL. This is not meant for serious use, but may be of
interest to people who want to play around with Dylan. We are using it
internally for cross-development and testing of ideas. We hope to have a
stand-alone Dylan implementation ready for release by Fall of 1994, though
this depends on a number of factors not entirely within our control. All
releases will be announced on "comp.lang.dylan".
Regarding intellectual property issues, our goals are to make the software
we produce easily available to the research community and to help
popularize Dylan, while still retaining some control over commercial use of
the Igor development environment and derivative works. Our tentative plan
is to copyright the Igor code, but with blanket permission to use the
library and runtime code. This should eliminate any concerns over "runtime
licenses" for commercial applications developed in Igor. Our Dylan
compiler and the Igor environment code will be copyrighted but free for
non-commercial use; anyone who wants to sell this code or incorporate any
of it into a commercial product will need a license from CMU. We plan to
set up a consortium for companies who are interested in working with us on
this effort. The details of that are still being worked out.
One final note: While we plan to base Igor on the Dylan language, we do not
yet have a license to use the name "Dylan", which is an Apple trademark.
We believe that this will be fixed by the time we have something ready for
general release.
-- Scott
===========================================================================
Scott E. Fahlman Internet: ····@cs.cmu.edu
Senior Research Scientist Phone: 412 268-2575
School of Computer Science Fax: 412 681-5739
Carnegie Mellon University Latitude: 40:26:33 N
5000 Forbes Avenue Longitude: 79:56:48 W
Pittsburgh, PA 15213
===========================================================================