home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
ftp.cs.arizona.edu
/
ftp.cs.arizona.edu.tar
/
ftp.cs.arizona.edu
/
icon
/
newsgrp
/
group01a.txt
/
000085_icon-group-sender _Wed Jul 5 08:03:33 2000.msg
< prev
next >
Wrap
Internet Message Format
|
2002-01-03
|
2KB
Return-Path: <icon-group-sender>
Received: (from root@localhost)
by baskerville.CS.Arizona.EDU (8.9.1a/8.9.1) id IAA24576
for icon-group-addresses; Wed, 5 Jul 2000 08:02:11 -0700 (MST)
Message-Id: <200007051502.IAA24576@baskerville.CS.Arizona.EDU>
From: Atle <trollet@skynet.be>
X-Newsgroups: comp.lang.icon
Subject: Re: Error messages
Date: Sat, 01 Jul 2000 12:21:15 -0100
X-Trace: news0.skynet.be 962446543 8139 194.78.236.172 (1 Jul 2000 10:15:43 GMT)
X-Complaints-To: abuse@skynet.be
X-Accept-Language: en
To: icon-group@optima.CS.Arizona.EDU
Errors-To: icon-group-errors@optima.CS.Arizona.EDU
Status: RO
Content-Length: 1414
Steve Wampler wrote:
>
>
> set := table()
>
> and then the line
>
> foo := set[bar]
>
> makes complete sense. (This type of language flexibility does
> make it difficult to produce error messages that always make
> sense to the user, who may be thinking of something entirely
> different - which is why having a traceback that includes line
> numbers is so useful!)
This may not be related to Icon at all (although I feel that Icon, ProLog, Simula and Beta are 'close cousins :) - but I have been
fooling around with the idea of writing a parser in ProLog - I started out with something close to the grammar for 8086 assembly.
Why? Because the syntax is simple, so syntax errors wont contain much information as to what the error actually is, but the
semantics possibilities are enormous compared to, say, Pascal.
Now, this is where I think it might be relevant: Wouldn't some kind of knowledge base help in producing more meaningful error
messages? I have been through Icon enough to write small programs in it, and I had the same impression: The error messages really
only say: "Error" - and you have to find the rest out for yourself. The case here about the [] could be added to the knowledgebase,
so the next time, the error would be "I told myself I wouldn't use [] here!" :-)
Any thoughts? Should I drop the idea of using an AI language to build an assembler?
(I actually wanted to use Beta)