Newsgroups: comp.lang.scheme
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!tank.news.pipex.net!pipex!peer-news.britain.eu.net!newsfeed.ed.ac.uk!edcogsci!jeff
From: jeff@cogsci.ed.ac.uk (Jeff Dalton)
Subject: Re: scsh in scm and Scheme gui's
Message-ID: <DKvr14.2u6.0.macbeth@cogsci.ed.ac.uk>
Organization: Centre for Cognitive Science, Edinburgh, UK
References: <4bjskn$o6b@jive.cs.utexas.edu> <DKo6wD.Lox.0.macbeth@cogsci.ed.ac.uk> <4cjsqm$ei@jive.cs.utexas.edu>
Date: Mon, 8 Jan 1996 20:46:16 GMT
Lines: 169

In article <4cjsqm$ei@jive.cs.utexas.edu> wilson@cs.utexas.edu (Paul Wilson) writes:
>In article <DKo6wD.Lox.0.macbeth@cogsci.ed.ac.uk>,
>Jeff Dalton <jeff@cogsci.ed.ac.uk> wrote:
>>In article <4bjskn$o6b@jive.cs.utexas.edu> wilson@cs.utexas.edu (Paul Wilson) writes:
>>
>>>>Quoth Olin, in the article to which Paul followed up:
>>>>
>>>>o> Contrary to the popular claim, people are *not* afraid of parens.
>>>>
>>>>Without wishing to make this sound like a flame, I find it somewhat
>>>>hard to believe that people are still willing to make such an
>>>>assertion in light of the treacle pace with which the Lisp industry
>>>>grows relative to the rest of the software industry.  Look at Franz;
>>>>they've shipped a good product for over a decade, and they're still a
>>>>tiny company.  Sure, AutoCAD is more of a success, but it's *one*
>>>>application.
>>>
>>>Agreed.  Look at Perl.  Look at TCL.  Look at Visual Basic.  These are
>>>the biggies, not AutoCAD.  While syntax isn't the only issue, it's
>>>not one that can be ignored. 
>>
>>Gee, maybe Olin has, like, _evidence_.  Maybe he's not just making
>>it up.  (I know of some evidence from Multics Emacs, but that was a
>>long time ago.  But I know there can be evidence.)
>
>Yes, and there can be evidence the other way.

There might be, if it's true that people are afraid of parens.

(Actually, I know there is some evidence.  A number of people _are_
out off by Lisp/Scheme syntax.  But it's not clear why this is so,
or how it might be overcome.  It may that in the right context,
people don't mind parens, that, properly introduced, the syntax is
not a big problem.  It may be that negative reactions to the syntax
are a compound of initial difficulty in reading it plus a general
impression that these languages are completely losing, that they're
slow, require vast amounts of memory, have failed totally in the
market, etc.  There's plenty of room in here for the possibility
that people are not afraid of parens.)

> I don't think Olin or Brian
>or I has *definitive proof*, but we all have some evidence.  I was just
>saying that the relative popularity of infix-but-ugly syntaxes is
>evidence.

But it's not unless there's evidence that the syntax was a significant
reason behind the relative popularity of the software in question.
Relative popularity of Perl, Tcl, etc. alone shows nothing about the 
syntax without some further argument or evidence.

_Now_, of course, you're talking in a different way.  _Now_ you're
saying "the relative popularity of ... syntaxes".  That, if it were
shown to be the case, would be evidence (given the role of parens in
Lisp/Scheme syntax).

But there are a number of reasons, other than syntax, why Perl, TCL,
etc might be more popular than Scheme.  I suspect that those reasons
are more significant.  They certainly are vastly more important in
my use of Perl and shells.  In any case, you can't just take the 
popularity of the software _as_ popularity of the syntax.

I note further the language of the exchange I was responding to,
in particular:

  I find it somewhat hard to believe that people are still willing to
  make such an assertion in light of the treacle pace with which the
  Lisp industry grows relative to the rest of the software industry.

This suggests that the "parens lose" conclusion follows directly from
the growth pace (which it does not), and that this is obvious, without
even considering that Olin might be in posession of more direct
evidence to the contrary.

>  If you take the popularity of AutoCAD and Emacs as evidence,
>you should also take the popularity of Visual BASIC, Tcl, and Perl
>as evidence.  This evidence takes some interpretation, of course,
>but it's evidence.

But the popularity of software that has a paren-filled syntax *does*
show that the syntax is not so such a barrier that it stops a number
of people from using the software.  The popularity of software with a
low-paren syntax *does not* show that parens are a barrier at all,
because other factors may be responsible.

It's true, of course, that the popularity of some software that has
a paren-filled syntax does not show that parens are _no_ barrier.
It just puts an upper bound (so to speak) on the barrier effect.

But, again, the popularity of low-paren software does justify a
symmetric conclusion: it does not show that that the barrier is at
least so high.  It doesn't, because it doesn't show that syntax
is a significant factor at all.

>There's other evidence I've got.  I've talked to a fair number of
>graphics people about interactive programming languages, and they
>tend to resist anything that looks like Lisp.  Maybe they're crazy,
>but they do.  I think that if I could give them my two-hour "Intro
>to Scheme" lecture, they'd be a lot less resistant.  (My students
>generally don't have a problem with Scheme, because they have to
>sit through an explanation of its beauty.)

I take your point, but you haven't said why they resist anything
that looks like Lisp.

>The "market" for Scheme is pretty limited at present.  It's largely
>an academic language, rather than a real-world language.  Part of
>the problem is that Scheme is just too small for a lot of real-world
>programs---people end up using a lot of nonportable extensions.
>But I think another problem is that the syntax limits its acceptance
>among people who haven't been taught Scheme in a class.  If they're
>looking around for a language, they'll skip over Scheme because it
>looks funny.

That could well be the case.

>I'm not claiming that this is the only thing going on, or that I
>have solid proof that it's an important thing.  But I think there's
>reason to think it's true to some significant degree---and that the
>examples of AutoCAD and Emacs are not compelling.

They show something (see above), but not more.  But anyway
I haven't been making a case based on the success of AutoCAD or
Emacs.  My remark on Multics Emacs (not a very widely used bit
of software) was about some evidence that non-programmers would
start writing little Emacs extensions w/o thinking they were
programming (something often seen as too hard) or, evidently,
being put off by the parens.

>Another reason why I don't think AutoCAD and Emacs are compelling is
>that those are "killer apps" with enough appeal that they can overcome
>the resistance to Lisp syntax.

That's a good point, but you're assuming a resistance in AutoCAD and
Emacs users that has not been shown (at least not by any evidence
given here).

>On the other hand, an interactive shell with a terse syntax would
>be a nice Trojan horse.

Very true.  Indeed, I'm far from being opposed to such moves.

>>There are plenty of other differences
>>between Allegro CL and Perl, just for instance.  (I use Perl from
>>some things where I wouldn't use CL, BTW, and this has nothing to
>>do with syntax.  It has to do with the size of CL programs at
>>run-time and the lack of fast enough text-based file-processing.)
>>
>>Note that I'm not saying Lisp syntax isn't a problem.  Maybe it is.
>>All I'm doing is objecting to an unjustified inference.  If anyone
>>has some better evidence that syntax is (or isn't) a barrier, let's
>>have it.
>
>I concede that I overstated my case---I should have made it clearer
>that what I was saying was "as I see it" and why.  Others can make
>their own best guesses as to what the important issues are.  I could
>certainly be wrong.

Well, I don't think we really disagree about very much here.

What I object to is the seemingly automatic and direct inference
from market share to syntax.  I think that, if we want Scheme to
be more successful, we need to look beneath appearances and
question assumptions, especially when our inferences match those
of people who've already decided, often without much evidence,
that Scheme is hopelessly losing.  If we don't question such
inferences, no one else will.

-- jd

