standardization

[MATHIS@USC-ISIF.ARPA]

I'm glad that we agree on such controversial issues as what we're
talking about: Lisp.

Scott is right, we should let people know what we're  doing.   He
and  I  are  meeting on Monday the 17th.  If we have any thoughts
from the rest of you before that, then he and I could  coordinate
a message.

I  intend to submit our proposal to X3/SPARC on the 18th.  Here's
the current version:


1  Identification of Proposed Project

1.1  Title -- Common Lisp


1.2  Proposer(s)

Robert F. Mathis, 9712 Ceralene Drive, Fairfax, VA 22032,
(703)425-5923, on behalf of the Common Lisp Community.

The Common Lisp Community is an informal collection of people
from industry, academia, and government who have particpated in
the initial design and implementation of Common Lisp.  This group
has been in existence for five years, communicating primarily by
ARPAnet.  Occasionally the community meets, as they did in
Boston, December 9-11, 1985.  At that meeting they decided that a
national and international standards effort for Common Lisp was
appropriate and endorsed Bob Mathis as the coordinator and
organizer of that effort.


1.3  Date Submitted -- February 18, 1986


2  Justification of Proposed Standard

2.1  Needs

Lisp is the second oldest programming language still in current
use (after Fortran).  Lisp has traditionally been the language
used for most Artificial Intelligence programming, and is now
becoming popular for non-AI tasks as well.  Throughout its early
history, Lisp was the subject of much experimentation; this has
greatly improved the Lisp language, but has also led to a
proliferation of incompatible dialects.  This lack of
standardization has impeded the acceptance of Lisp in industry.

In 1981, with the encouragement of DARPA, an effort was begun by
a number of researchers at several organizations to define a
commonly acceptable version of Lisp.  The language specification
was written by members of this informal group, after extensive
discussions on the ARPAnet.  The resulting book, Common Lisp: The
Language by Guy Steele Jr., has received world wide acceptance.
MACLISP, ZETALISP, SCHEME, INTERLISP, SPICE LISP, S-1 LISP, NIL
(New Implementation LISP), "Standard" LISP, and Portable
"Standard" LISP have all been considered in the design of Common
Lisp; and the most useful features of each were incorporated.
Common Lisp, as described in the Steele book, has now become a de
facto standard within the US, and there is great interest in this
language in Europe and Japan.  Therefore, we feel that the time
has come to develop an official national and international
standard for this language.


2.2  Recommended Scope of Standard

The scope of the proposed standards effort is essentially the
same as the scope of the Steele book. Because of the nature of
the language and its implementations, the distinctions between
implemented language features, predefined system functionality,
and user defined supplementary capabilities are not the same as
in other languages; but there will still be some issues to
resolve about the size of the language, possible subsets and
supersets, and implementors' options.


2.3  Existing Practice in Area of Proposed Standard

Of the US hardware manufacturers who support any kind of Lisp on
their machines, the overwhelming majority have now announced
plans to implement the Common Lisp language as described in the
Steele book, either as their only supported Lisp product or
alongside some older dialect.  DARPA and other agencies in the
DoD are insisting on Common Lisp as the version of Lisp to be
used on their projects.  A Common Lisp Committee has been formed
in Japan to promote the language there. There is growing interest
in the language in Europe.


2.4  Expected Stability of Proposed Standard with Respect to Current

Since the Common Lisp language is still relatively new, we
anticipate that new revisions of the standard would appear more
frequently than for other standardized languages (approximately
every two or three years at the beginning).


3  Description of Proposed Project

3.1  Definitions of Concepts and Special Terms (if any)

None at this level and time.


3.2  Expected Relationship with Approved X3 Reference Models

None at this time.


3.3  Recommended Program of Work

3.3.1  Base Documents

Common Lisp: The Language by Guy Steele Jr., Digital Press,
Burlington, MA, 1984.


3.3.2  Time/Milestone Schedule

Preliminary draft plan of work for ANSI/X3 standarization of
Common Lisp with the formation of a new technical committee
(probably X3J13).

Milestone 0
February 18, 1986   These Documents to SPARC

Milestone 1 & 2
March 18-20, 1986   SPARC Meeting

Milestone 5
March 27, 1986      X3 Ballot out for "30 days"
May 2, 1986         X3 Ballot closes
May 27, 1986        Status Resolved
May 28, 1986        Press Release & Meeting Announcement
(August 4-6, 1986   ACM Lisp Meeting in Boston)

Milestone 6
September, 1986     X3J13 First Meeting
November, 1986      X3J13 Second Meeting
November, 1986      X3 Ballot on Officers
December 30, 1986   X3J13 Organization Complete
March, 1987         X3J13 Meeting on Issues
May, 1987           X3J13 Meeting on Drafting

Milestone 7
July, 1987          X3J13 Meeting on Draft

Milestone 8
Aug-Oct, 1987       X3J13 Ballot on Draft

Milestone 9
November, 1987      X3J13 Resolution of Issues

Milestone 10
January, 1988       Submission to X3

Summer, 1988        X3 Ballott complete
Fall, 1988          ANSI Standard Common Lisp


3.3.3  Potential Participants

Some leaders in the Common Lisp Community have already been
identified who are willing to serve: John McCarthy, Stanford
(inventor of LISP); Guy Steele, Thinking Machines (author of
Common Lisp: The Language); Scott Fahlman, CMU; Dick Gabriel,
Lucid; Dave Moon, Symbolics; Steve Squires, DARPA; Ron Ohlander,
USC-ISI; and Bob Mathis, Private Consultant. Other people we also
expect to be involved include (this is only a partial list):
Danny Bobrow, Xerox; Martin Griss, H-P; Jonathan Rees, MIT;
Jerome Chailloux, INRIA; and Masayuki Ida, Aoyama Gakuin
University.


3.4  Resources -- Individuals and Organizations Competent in Subject

Some of the individuals have been identified in Section 3.3.3
above. We also expect participation from major companies and
universities.


3.5  Recommended X3 Development Technical Committee (Existing or New)

We are recommending the formation of a new X3 Technical
Committee, which will probably be known as X3J13, for the work on
Common Lisp.


3.6  Anticipated Frequency and Duration of Meetings

Two meetings of two days each in 1986. Four meetings of two days
each with considerable additional discussion over the ARPAnet and
other correspondance media in 1987. Three or four meetings per
year after that.

A more frequent meeting schedule is planned at the beginning to
develop the standard we are all waiting for. After that the group
will have learned how to work together ina standards context and
meetings will only be necessary to confirm discussions and
tentative resolutions achieved by correspondance (primarily
electronic).


3.7  Target Date for dpANS to X3 (Milestone 10)  - January, 1988


3.8  Estimated Useful Life of Standard  -- 25 Years (with revisions)


4  Implementation Impacts

4.1  Impact on Existing User Practices and Investments

The Common Lisp design was based on the most popular dialects of
Lisp. The difficultly of conversion from one dialect of a
language to another depends not only on the dialects but also the
programming style of the user. Conversion to Common Lisp seems to
be in the same direction as current Lisp style and philosophy.


4.2  Impact on Supplier Products and Support

As mentioned earlier, of the U. S. hardware manufacturers who
support any kind of Lisp on their machines, the overwhelming
majority have now announced plans to implement the Common Lisp
language as described in the Steele book, either as their only
supported Lisp product or alongside some older dialect.  DARPA
and other agencies in the DoD are insisting on Common Lisp as the
version of Lisp to be used on their projects.


4.3  Techniques and Costs for Compliance Verification

The implementors in the Common Lisp community agreed at the
December 1985 meeting to begin sharing their existing
implementation test suites.  USC-ISI will collect them and make
them available on-line to interested parties.  Proposals for a
more comprehensive validation or certification process are now
being discussed.


4.4  Legal Considerations

The Steele book was published by Digital Press. Guy Steele has
already contacted them to be sure that no problems will arise
because of their copyright. It is important that the text of the
standard also be available in machine readable form for users and
implementors.


5  Closely Related Standards Activities

5.1  Existing Standards -- NONE


5.2  X3 Standards Development Projects -- NONE


5.3  X3/SPARC Study Groups -- NONE


5.4  Other Related Domestic Standards Efforts -- NONE


5.5  ISO Standards Development Projects
At the first plenary meeting of ISO/TC97/SC22 in November, 1986,
in Paris, an ad hoc committee on "LISP, PROLOG, and Other
Artificial Intelligence Oriented Languages" was established and
Bob Mathis was named as its convenor. The Common Lisp Community
continues to be in close touch with that ad hoc committee's work.

The Common Lisp Community has already expressed a desire to have
an international standard and would like the US member body
(ANSI) to propose an international work item for Common Lisp and
also offer to have the secretariat. The SC22 TAG has already
considered this idea and endorsed Bob Mathis as a potential
convenor of an ISO working group on Common Lisp.

A New Work Item (NWI) proposal for X3 to forward to ANSI and on
to ISO for work on Common Lisp is being submitted concurrently
with this proposal for the establishment of a new X3 project on
Common Lisp.


5.6  Other Related International Standards Development Projects

There is a group in Europe working on a project they call EuLISP.
They have not yet taken their work to the standards organizations
in their home countries. It includes members from France, UK,
Germany, and possibly other countries.  They characterize their
work as drawing on the Common Lisp experience.


5.7  Recommendations for Close Liaison

The ISO/TC97/SC22 Working Group (if one is formed), the EuLISP
group mentioned above, the informal Common Lisp Community, and to
a lesser extent the various professional groups interested in
artificial intelligence applications and programming.