[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

ISO News

I want to briefly report what happened at the ISO meeting in Sendai,
Japan.  The reason is that some other folks I've talked to about it claim
that I need to be careful about how I report it to X3J13, and I want to
see if you have any remarks.

The following drafts were submitted:

	Common Lisp
	LeLisp (Version 16, but represented by Version 15.22 with a list of
	        major changes.)

The German work plan was rejected (that is, the plan to accept the above
drafts, evaluate them, and select some for standardization) as part of a
complicated preference voting scheme (proposed by the Germans) on the list
of alternatives listed later in this message.

Apparently commercial concerns in the UK convinced BSI to convince Padgett
to allow short term (at least) use of CL in the UK, so he proposed EuLisp
for the long term only. Padgett stated that an ANSI standard without a
conflicting ISO standard would accomplish the goals of the UK.

Siemens (that is, DIN) stated they intend to follow ANSI Common Lisp.

Japan requested that X3J13 consider whether an ISO Lisp which is ANSI
Common Lisp without CLOS would be possible and to prepare the draft for it.

AFNOR proposed that an international committee be set up to determine a CL
subset for ISO.  The interesting thing is the AFNOR definition of subset:
L1 is a subset of L2 iff there exists a program (file full of stuff) c in
L2 such that for every valid program p in L1, the program p has the same
observable side effects in an implementation of L1 as the program c
concatenated with p has in an implementation of L2.

AFNOR refers to c as a compatibility program, but looking at the stuff they want
in the subset, I think c needs to be a compiler.

By the way, Chailloux, who is on leave from ILOG and working at DEC-SRC in
Palo Alto in the Modula-3 group, did not attend because his father died
the day before he was to leave for Japan.

The list of alternatives and their rankings are as follows:

1.	abandon the goal of a short term standard and start working on the long

2.	ask X3J13 to consider CL-CLOS and to prepare that draft (Japan
	proposed this).

2. 	Form an international subcommittee of WG16 to define a subset of CL

4. 	Wait two years and try the German plan again.

4.	Use EuLisp and the KL draft (a Kernel Lisp the Japanese are designing
	and of which a draft was submitted for information only this time.
	It is a not-so-bad subset of CL with some changes to packages.) as the
	starting place while doing the long term.

6. 	Continue with the German plan.

The Japanese were very unhappy with dropping the short term, so I backed
them in walloping the convenor until he agreed it wasn't the consensus.
This means we were left with the two alternatives tied for number 2, but
AFNOR blocked having WG16 ask X3J13 this question, so the Japanese and the
US made it clear that the US was happy to accept the private Japanese 

The action items are for each country to submit a list of the features of
CL required in the subset and a list of those things it wants to drop, and
for each country to list starting places or criteria for the long term. The
next meeting is Feb 1-2 or Feb 2-3 in Palo Alto or San Jose.

About half a day was devoted to commenting on the drafts. The interesting
discussion on the CL draft was that it now clearly defined the ambiguities
rather than vaguely defining them. One example is that the evaluation
model states clearly that the functional position might be evaluated
before all arguments or after all arguments.

The question I have is whether to press hard with the CL-CLOS proposal.
The Japanese commented privately that CLOS would probably be adopted by
Japanese implementors, but that they were insistent that an ISO standard
should enable existing programs to run in preference to providing a
language to write new ones.  They suggested that `later' we could consider
adding CLOS, if it was a commercial success.

An important step forward was that they dropped the idea that CL = CLtL and
accepted that CL = X3J13 CL.

I believe that if we produce CL-CLOS, the following countries will vote for it:

	US, UK, Germany, Italy, Japan.

France will vote against it.

I don't know how Denmark or The Netherlands will vote, but I think they
will vote for it. If Switzerland and Canada still vote, they will vote for
it. I suspect that a savvy AFNOR delegation can stall forever, as can we.

Gregor believes that we can eventually ratchet Japan from CL-CLOS to CL,
and that I should not push X3J13 to wholeheartedly comply with Japan, but to
comply some other way (Quux, we can talk about that separately).

What do you think?