Newsgroups: comp.lang.lisp
Path: cantaloupe.srv.cs.cmu.edu!das-news.harvard.edu!news2.near.net!MathWorks.Com!europa.eng.gtefsd.com!howland.reston.ans.net!EU.net!uknet!festival!edcogsci!jeff
From: jeff@aiai.ed.ac.uk (Jeff Dalton)
Subject: Re: Lisp advocacy (Was Re: another take on "C is faster than lisp")
Message-ID: <CwFvtH.LAt@cogsci.ed.ac.uk>
Sender: usenet@cogsci.ed.ac.uk (C News Software)
Nntp-Posting-Host: bute.aiai.ed.ac.uk
Organization: AIAI, University of Edinburgh, Scotland
References: <779114796snz@wildcard.demon.co.uk> <Cw3123.6Cs@cogsci.ed.ac.uk> <779659921snz@wildcard.demon.co.uk>
Date: Tue, 20 Sep 1994 17:41:40 GMT
Lines: 98

In article <779659921snz@wildcard.demon.co.uk> cyber_surfer@wildcard.demon.co.uk writes:
>In article <Cw3123.6Cs@cogsci.ed.ac.uk> jeff@aiai.ed.ac.uk "Jeff Dalton" writes:
>
>> The views of the immediate participants typically don't change
>> right away, but they're not the only people who matter.
>
>Is that a justification of endless "X vs Y" debates? 

No, it's just a reply to one point against such debates.

>      I agree with
>you about their views (if I understand you correctly), but I don't
>see the value in everything I see in these debates. They look like
>at least a few of participents are not looking for a constructive
>result. I sense a desire for a _destructive_ result, like deminishing
>interest language X or language Y.

I think that's true in some cases, though it's difficult to be sure
about motives.  But I can say something about my case.

Perhaps my position is unusual, and hence easily mistaken for
something else.  I don't know.  But I don't say Lisp is better
than X, Y, or Z.  (Ok, I might in extreme cases, but not for
languages like C or Beta.)  Instead, I try to show that it can
be reasonable to prefer Lisp and why someone might to so, and
I try to keep open some possibilities that are in fact open.
(By "open" I mean that so far as we know they are possible and 
not impossible, hence not "closed".)  This does not strike me 
as a "religious" position as that is usually defined.

I'm probably not very effective as a Lisp "advocate".  I'm too
annoying, and I write too much.  But I'm after a constructive
result along two fronts:

  1. That Comp.lang.lisp present a true picture of Lisp.

The true picture does not include "Lisp is better than C", or 
anything like that, but it does include the full complexity of Lisp:
that it's a family of languages that differ in many ways and for 
which implementations vary much more than, say, C implementations; 
and that a number of implementation strategies and techniques have not
yet been fully explored.  (Indeed, some are only just starting to be
explored.)

  2. That complaints about Lisp be sufficiently specific that 
     people reading comp.lang.lisp can tell (a) how serious the
     problem is and (b) what are the possibilities for fixing
     the problem by changing the language or the implementation.

For (b) I want something specific, such as "change the buffer
size to 1 K instead of 2" or "this GC excessively reduces
locality of reference because ..., but this other strategy is
better" or "optimize leaf procedures in following way".
(See the discussion of "data bloat" for another example.)

In some cases, we won't know the full story -- maybe we don't know
whether the problem can be fixed, or maybe we don't know exactly
why it occurs -- but this restricts the conclusions we can draw.
If we don't know whether it's possible to completely remove the
problem, then we *don't know*.

>These debates are often called "religious wars". I'm questioning the
>constructive value of such wars, and the appropriate place to hold
>them. 

My understanding of "religious wars" was that they involved
opinions presented as facts.

In any case, I have little interest in arguing whether Lisp is
better or worse than something else.  I do have some interest
in people reaching conclusions about Lisp (and about particular
Lisp-family languages and implementations) for good reasons
rather than bad ones.  But a good reason could be something
like "I can program well in C and I know it will be suffficiently
efficient" (if, of course, C really can be suffficiently efficient).

>I'm also assuming that comp.lang.lisp is a newsgroup for
>constructive debates. Perhaps this is an impossible expectation,
>I don't know. I've been using UseNet for about 2 years, so I'm
>sure I'm not experienced enough to judge it in this respect. Plus,
>I've seen similar debates elsewhere, with what appears to me to be
>an equal lack constructive discussion, compared with discussions
>that don't compare two or more languages.

I thought the initial postings on Beta vs Lisp were fairly useless,
because they didn't say enough for me to tell why certain remarks
were being made.  (E.g. why threads would make WITH-HASH-TABLE-
ITERATOR obsolete -- which we straightened out vai e-mail -- or
what the "incredible symmetry" was.)

>Thanks for your contribution to this debate. I'm just a little
>puzzled by it. Could you please help me? Perhaps I've misunderstood
>your post.

Does the above help at all?

-- jeff

