Paulo J. Matos wrote:
> This is my first post to this ML but I'm following the discussion
> since the beginning and I can't understand your reasons at all.
I've enclosed a guide that might help.
> [...] what you want [....] not only seems anarchic to me but also
> defies the increased portability R6RS is about to bring.
Contents
Summary of Conclusions
Summary of Argument
Final Remarks About Authority
Summary of Conclusions:
a) the best outcome would be voluntary attrition from the
process by both the steering committee and editors, with
the intention that they re-convene around a SRFI
b) in the alternative, people should boycott the vote
because it is illegitimate
c) in the alternative, people should submit a vote
of "abstain" with a rationale that spells out the
illegitimacy of the process
d) in the alternative, people should vote "no", with a
rationale that public discussion of the largest
changes in the draft has been inadequate
e) in the alternative, people should vote "no",
with a rationale that explains objections to
some of the large changes in the draft
and
f) should the process complete and end without
ratification, the draft should be submitted as
a SRFI
g) should the process complete and end with
ratification, that will be a damn shame
Summary of Argument:
1. There is no *need* to publish as R6RS
1a. because the portability outcome will be the same
(It won't change the plans of current implementers
much. We don't know what future implementers will
do either way.)
1b. because the formality of the document will be the same
(It's the same words on the page, the same history of
how it was created, and the same "signatures", either
way.)
1c. because the economic value of the document will be
at least as good
(A SRFI, like an R6, can be cited in contracts,
grant applications, text books -- any economically
significant place where a reference to a formal
standard is needed, a SRFI will do, for all
practical and principled purposes. *This* SRFI
can (and probably should) include "signatures"
and endorsements from a very large number of
people, distinguishing it clearly from less important
SRFIs.)
2. There are *reasons to not* publish as R6RS
2a. because the report process is too slow
(A process which has taken this long since
conception to nearing ratification is too
slow to operate in meaningful feedback with
the Scheme community.)
2b. because this report process is too divisive
(Most of the major decisions in the draft
were taken in private, among the implementers.
The formal comments have succeeded in collecting
"bug fixes" to the draft and at opening up
some "what color shall we paint the bike shed"
issues (like library version numbers), however,
notably, all formal comments attempting to
revisit some of the major decisions have been
rejected and with very little meaningful two-way
discussion or debate.)
2c. because the authority to publish revised reports
no longer exists, with anyone
(The draft demonstrates that, if nothing else,
the historic continuity of the revised reports
-- the continuity of a line of thought that
is rooted back the "lambda the ultimate" papers --
has broken. People have moved away from the
process and new agendas have moved in.
People who have been with the report process
all along are in very, very different places
than they were in revisions past. From
a literary perspective, if nothing else,
there would be some kindness in allowing R5
to be the last revised report, ever.)
2d. because the R6 process is anti-democratic
(Formally, the R6 process is tyrannical.
A sui generis oligarchy has granted itself
sovereignty over community membership (the
150 word essay issue), monopoly to publish
draft "laws" for possible adoption,
and the authority to override any vote,
even of the "approved" citizens. No
rationale has been offered for this arrangement
other than the definite economic goals of
a few and the alleged economic benefits to
others. The R6 process is basically of a
tyrannical oligarchy offering a (pretty thin,
pretty theoretical) bribe (of a world rich in
portable Scheme code that could not possibly
be achieved without the oligarchic assertion
of sovereignty) to collect votes which
give a retrospective appearance of legitimacy.)
2e. because revised reports cause terminological
confusion and impose unfair economic costs
on dissenters
(I am sure I'm not alone in wondering, should
R6 pass, what I will call the implementation
I'm working on since the proper name "Scheme"
will have been so radically redefined.)
2f. because this draft in particular piles on
many complex features yet no rationale has
been offered that shows they can not be usefully
reduced to a few more general, less restricted
constructs
(An example: the draft's Unicode support could
have been enabled mostly by removing a few
restrictions in R5, making the CHAR type
more general. Instead, the draft proposes
some hefty new restrictions on the CHAR type.)
3. There are *reasons to* publish as a SRFI
3a. because the SRFI process is relatively fast
so that later corrections, amplifications,
and alternatives to the draft can be published
along side the draft in a timely manner
(This is inherent in the SRFI process.)
3b. because the SRFI process is relatively non-rival:
it does not propose to define the borders of
"the Scheme community"
(This is inherent in the SRFI process.)
3c. because the authority to publish a SRFI is available
(This is inherent in the SRFI process.)
3d. because the SRFI process is democratic
(This is only approximately true but it is a
good approximation, inherent in the SRFI
process. There is still a (formally) tyrannical
oligarchic imposition of minimum standards of
relevance, language, form, etc. however (a) those
are relatively objective concerns so it is easy
to see if they are being administered well;
(b) should the "SRFI oligarchy" go insane
and abuse that authority, it will be trivially
easy to "fork" the SRFI process without much
loss.)
3e. because SRFI's provide a non-rival, structured
name-space for standards rather than (further)
muddying the meaning of the proper name "Scheme"
(The literary function of the revised
report series is to offer an authoritative
definition of, to quote the title, "The Algorithmic
Programming Language Scheme".
I propose that that literary function is
obsolete. In ordinary usage, the proper name
"Scheme" is the name of a loosely defined set
of design patterns for programming languages,
where all programs that conform to the patterns
contain a common, portable, tiny subset in
terms of which all else in the language can
(in theory) be defined. The report series
has long aimed to define that subset, and
the current draft diverges from that tradition
substantially.
It is surely useful to have a variety of
formal documents, the product of public debate,
to which to refer as we do and promote our work
in Scheme. The SRFI process provides us with
a non-rival name-space for such documents.)
Final Remarks About Authority
Paulo J. Matos also wrote:
> You can vote no to the ratification! What more do you want? There
> will always have to be some group at the top making the hard
> decisions. If you don't like them, vote no but putting all the
> hard decisions in the hand of a community is not wise, IMO.
What you are saying, in effect, is that the Scheme community
is populated by a bunch of superstitious idiots for whom
the magical title "R6RS" will hold some important totemic
significance not found in a title like "SRFI-<N>: First Report
on the Scheme Workshop Programming Environment".
I have higher expectations of this community than you do,
I think, although I admit I have fears about equally as low.
-t
Received on Fri Jun 15 2007 - 13:54:56 UTC
This archive was generated by hypermail 2.3.0
: Wed Oct 23 2024 - 09:15:01 UTC