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

Re: GNU Extension Language Plans



> As you all probably know, Richard Stallman announced plans to base a
> GNU Extension Language on Scheme.

This is not quite accurate. He is basing his language on a specific
implementation of Scheme, namely SCM. I believe he has made this
decision based on incomplete information.

> Given the broad support for Scheme
> promised, I suggest that the Scheme Authors some how show support for
> Stallman's effort.  Maybe the group should post an endorsement of
> GEL on gnu.announce.  What do you think we should do to help?

I don't think it is our place, as a group, to engage in "endorsement"
of products any more than we already have by having nurtured a
well-crafted programming language for public use.

More importantly, I would sincerely like to see Stallman's efforts in
this direction succeed but I would like him to be better informed
about the issues involved. Frankly, I believe he has an outdated view
on the state of the Scheme language as well as the state of its
various implementations and, consequently, he is likely to encounter
problems he does not anticipate if he over-commits himself too early in
the design process to a specific implementation of the language.

I do not intend this to disparage SCM as an implementation. IT serves
a useful purpose and enjoys a deserved place in the implementation
spectrum. I also do not mean to call into question Stallman's
competence as a software engineer. His technical reputation speaks for
itself. I mean only to call into question the advisability of
Stallman's premature committment. My office is only a half dozen doors
down the hall from Stallman's. I overheard a critical turningpoint in
his decision of which implementation to use.  He was clearly not
completely informed at that point. Specifically, he was primarily
concerned about the license agreements of various implementations and
admittedly he knew little about many of them, although he conjectured
that he could imagine what they might be. He seemed to chose SCM
almost exclusively because it was already distributed by GNU and
therefore, I believe, compatible with the GNU public license. (MIT
Scheme, the only other Scheme implementation that I know of which is
distributed by GNU, is presumably unsuitable as a light-weight
scripting language implementation as it is aimed at more high-powered
performance-critical large software systems. Presumably.)

I would therefore assert that he has made a decision to use Scheme
based on technical merit but he has specifically chosen SCM based
largely on incomplete information and personal political motivations.

I believe that our position in this situation should not be one of
endorsement but one of education. Specifically, I think we would save
Stallman a lot of trouble and effort by offering him expert advise
about the various Scheme implementations as well as their various
license agreements.

More importantly, I believe he should be reminded that it is unwise to
anchor his efforts in any specific implementation.  Rather, he should
attempt, whenever possible, to make his efforts portable across a
range of Scheme implementations. He should also enter into a direct
dialog with the community of Scheme implementors so that we might
better serve his needs or at least come to understand the design
pressures he deems most critical so that we might improve the Scheme
language as a whole.

I believe that if he conducts his efforts in this more liberal way, he
can serve to motivate growth and progress in the Scheme community by
cogently inviting various implementors to provide the features he
deems important. In this way, we can all benefit, as a community, from
his input. It could also serve to help the Revised 5th Report efforts
by re-vitalizing _our_ committment to providing an increasingly useful
and relevant on-going communal language design effort. Participating
directly in a real-world engineering effort in this way could be a
very healthy and productive enterprise.

If, on the other hand, he stubbornly insists on committing to one
particular implementation based primarily on personal political aims,
then I believe, whether or not he is successful, his efforts will
serve to stifle progress in Scheme. As one member of the RNRS list has
already gloomily predicted, this would pretty much spell the end to
Scheme as a living language. More precisely, it was worded "Have you
seen Stallman's Scheme obituary on the Scheme list yet?"

For this reason, I feel that an endorsement of his project, although
well intentioned, is inappropriate at this time. I suggest that,
instead, we try, individually or as a group, to temper his rash public
announcement in an effort to encourage him to embrace _our_ agenda of
fostering a variety of alternative Scheme implementations in order to
support a wide range of experimentation within the Scheme language
framework.

In this way, a tragically selfish and ignorant kidnap/assassination of
Scheme can be avoided, whether or not it is intentional or accidental.

Note, by the way, that Stallman does not himself read Usenet news
groups or the Scheme Digest electronic mail redistribution. The best
way to contact him is probably by personal electronic mail or by
telephone. He is rms@gnu.ai.mit.edu. Finger him there for his phone
number. I don't know, off hand, what it is.

------------------------------------------------------------------------------
 ziggy@ai.mit.edu    Michael R. Blair   MIT Artificial Intelligence Laboratory
 (617) 253-0765 [O]        -.           545 Technology Square --- Room 439
 (617) 661-3394 [H]         /\.         Cambridge, MA   02139-3539