Newsgroups: comp.lang.dylan
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!newsfeed.internetmci.com!howland.reston.ans.net!ix.netcom.com!netcom.com!NewsWatcher!user
From: demars@netcom.com (Dennis D.)
Subject: Re: dot syntax confusion
Message-ID: <demars-1502962132020001@10.0.2.15>
Sender: demars@netcom17.netcom.com
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
X-Newsreader: Yet Another NewsWatcher 2.0.3b0
References: <9602031834.aa29300@ax433.mclink.it> <demars-1302962053390001@10.0.2.15> <4g06fa$q3t@cantaloupe.srv.cs.cmu.edu>
Date: Fri, 16 Feb 1996 05:32:02 GMT
Lines: 46

In article <4g06fa$q3t@cantaloupe.srv.cs.cmu.edu>, nkramer@cs.cmu.edu
(Nick Kramer) wrote:

>Dennis D. <demars@netcom.com> wrote:
>>I don't think the problem is the dot notation, but the decision to allow
>>"-" in indentifiers. 
>
>The problem isn't that the Dylan designers loved hyphens.  They don't.
>The problem is that there's nothing better out there for forming
>multi-word identifiers.  Other choices are underscored_identifiers and
>MixedCapsIdentifiers, both of which evoke strong hatred in some
>crowds.  (Personally, I would have chosen underscores anyway, but
>that's a different matter)  Anyhow, I hope that when the ultimate
>development environment comes along, it'll give the user the option of
>displaying multi-word identifiers in different ways.  Perhaps it won't
>even use ASCII characters, but instead some kind of dot.
>

Actually, there was one language that had the perfect solution to this
problem: Algol 68, which permitted _spaces_ in varable names! (I remember
when a prof in a class I took at UCLA wrote some Algol 68 code on the
blackboard he observerd "You're won't see this in any other computer
language." Twenty years later and he's still right). Actually, it is a
little observed fact that Fortran allowed this too, but as most Fortran
compilers only allowed six characters or so for variables this was mostly
of academic interest.

Algol 68 had some other interesting typographical devices for improving
readability too -- like requiring keywords and some other lexical items to
be in boldface (and that in the days when most programmers were still
writing code on punched cards).

>>What should have been done (IMO) is to exclude this (and probably all
>>other arithmetic operators) from the identifier character set. 
>
>I've never heard a single person complain that the <type> naming
>convention was hard to read.  Although that possibility was also
>considered...
>-- 

Oh, OK, I was thinking of +,-,*,/ . I find <type> not only readable but
one of the better Dylan innovations (adds more readability than any other
Dylan convention I can think of). Since the context of where the angle
brackets are used for type identifiers is so different from the context of
arithmetic comparison, perhaps that's why it doesn't cause a problem.
- Dennis D.
