Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!cs.utexas.edu!yale!cmcl2!lanl!cochiti.lanl.gov!jlg
From: jlg@cochiti.lanl.gov (Jim Giles)
Newsgroups: comp.lang.fortran
Subject: Re: Throwing stones at standard committee (was Cheating on the types)
Message-ID: <19179@lanl.gov>
Date: 26 Mar 91 18:28:25 GMT
References: <1991Mar20.195732.15376@appmag.com> <10146@exodus.Eng.Sun.COM> <10208@exodus.Eng.Sun.COM> <8517@mentor.cc.purdue.edu> <1991Mar22.202121.26272@jato.jpl.nasa.gov> <18835@lanl.gov> <1991Mar26.005251.8162@jato.jpl.nasa.gov>
Sender: news@lanl.gov
Reply-To: jlg@cochiti.lanl.gov (Jim Giles)
Organization: Los Alamos National Laboratory
Lines: 72

In article <1991Mar26.005251.8162@jato.jpl.nasa.gov>,
vsnyder@jato.jpl.nasa.gov (Van Snyder) writes:
|> [...]
|> I didn't complain about complaining about the proposed standard. 

I'd like to complain about people who hold things up by complaining
about people complaining - something should be done about it.
                                    Monty Python
(Followed by a 16-ton weight being dropped on the man's head.)

|> [...]                                                     I complained
|> about asserting that the members of the standards committee were unable to
|> understand general principles of language design.

I didn't make such an assertion.  And neither did Herman Rubin (who
was legitimately pointing out considerations that standards committees
_do_ tend to consistently overlook).

However, now that you've brought it up: I'd _not_ claim that individual
members of the committee were unable to understand general principles of
language design, I _would_ claim that the committee as a whole is unable
to practice such principles.  Indeed, aside from ALGOL 60, I can't think
of a single successful language which was committee designed (and ALGOL
was _mostly_ done by a small subset of the committee).

Why should this be so?  Consider the answers to the first public
review.  Most consisted of "the committee considered your objection
but decided to do X anyway."  One of the committee members explained
such non-answers (in this very newsgroup) as being the result of the
fact that it was not possible to get a majority of the committee to
agree to a technical rationale for individual features.  It is my 
opinion that if there is not a single, easily described technical
reason for a new feature, it should not be included at all!  It 
appears that much of the new proposed standard consists of politically
"convenient" compromises.  That is: "you vote for my pet feature and
I'll vote for yours."  This is _not_ the way to design a language - or
even new features for one.  Committees should stick to what they do
fairly well: standardizing common practice.

|> [...]
|> >been as unacceptable as yours (or more so).  Complaining in a public
|> >forum is the only option left to those of us whose comments were
|>  ^^^^^
|> I didn't publish my criticisms -- X3J3 did.

Unfortunately, they didn't.  They just sent them around to committee
members and alternates.  It is my opinion that they _should_ be 
published as widely as possible in order that the user community
get the benefit of the thinking of other users.  The responses should
also be published (all of them) so the user community can see how the
committee has dealt with each issue.  

|> [...]                                                            But I
|> understand that in some environments, the tradition is first to try to
|> demonstrate that somebody with whom you disagree can't possibly possess the
|> mental and moral capacity to be right, or even sincere, and only then to
|> disagree with what he said.

There is another school of thought which expects honesty and
competence from others until such time as it is clearly demonstrated
not to exist.  The committee's responses to the public reviews as well
as other actions are a matter of public record.  I have bent over
backward in the assumption that their conduct is a result of political
deadlock, lack of time and low budget within the committee as a whole
and not a result of incompetence or dishonesty of individual members.
But, the fact remains, the net effect is unsatisfactory and deserving
of criticism.  It is all the more important that such discussion be
public since the vast majority of Fortran users are unaware of the
issues or have only heard the pro-Fortran 90 story from the available
books on the subject.

J. Giles