home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
ftp.cs.arizona.edu
/
ftp.cs.arizona.edu.tar
/
ftp.cs.arizona.edu
/
icon
/
newsgrp
/
group93a.txt
/
000032_icon-group-sender _Tue Jan 19 01:36:47 1993.msg
< prev
next >
Wrap
Internet Message Format
|
1993-04-21
|
1KB
Received: by cheltenham.cs.arizona.edu; Tue, 19 Jan 1993 05:48:11 MST
Date: Tue, 19 Jan 1993 01:36:47 MST
From: "Clint Jeffery" <cjeffery>
Message-Id: <199301190836.AA02412@chuckwalla.cs.arizona.edu>
To: icon-group@cs.arizona.edu
In-Reply-To: (Richard L. Goerwitz's message of 18 Jan 93 16:35:57 GMT <1993Jan18.163557.6566@midway.uchicago.edu>
Subject: detab/entab - anyone use them?
Status: R
Errors-To: icon-group-errors@cs.arizona.edu
>I like having entab/detab as functions in the language.
Yes, but couldn't they just as well be library functions, and not
part of the core language implementation?
The conversation about a minimal, extensible Icon is getting interesting.
A smaller Icon would have several compelling benefits for both users and
language implementors, but I don't think we can "rob Peter to pay Paul"--
so my question is: When is something in the language, but not in the language?
I think if Icon's linker knew about a standard archive of IPL procedures,
and pulled out those procedures whose names appear as (undeclared)
identifiers in the source program, we could make a library implementation
of entab/detab so easy to use that most folks would consider them built-in.
And I know a hundred people are going to jump on me for saying it, but
personally I think performance is not a very good rationale for entab/detab
to be built-ins -- convenience and backward compatibility are better arguments.