home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
ftp.cs.arizona.edu
/
ftp.cs.arizona.edu.tar
/
ftp.cs.arizona.edu
/
icon
/
newsgrp
/
group94b.txt
/
000059_icon-group-sender _Tue Sep 20 14:41:36 1994.msg
< prev
next >
Wrap
Internet Message Format
|
1995-02-09
|
2KB
Received: by cheltenham.cs.arizona.edu; Tue, 20 Sep 1994 08:32:44 MST
To: icon-group-l@cs.arizona.edu
Date: Tue, 20 Sep 1994 14:41:36 GMT
From: eddie@festival.ed.ac.uk (Eddie Corns)
Message-Id: <CwFnHC.Ds9@festival.ed.ac.uk>
Organization: Edinburgh University
Sender: icon-group-request@cs.arizona.edu
References: <1994Sep4.220207.10360@midway.uchicago.edu>, <34gaps$ste@ios.com>, <34ok80$nnb@lowell.bellcore.com>
Subject: Re: Icon - still alive??
Errors-To: icon-group-errors@cs.arizona.edu
norman@flaubert.bellcore.com (Norman Ramsey) writes:
>In article <34gaps$ste@ios.com>, Nick Williams <nmw@ios.com> wrote:
>>I just don't see what is wrong with standardizing an Icon library of
>>system specific functions. If standard would be welcome I would
>>volunteer time to write/implement it.
>I would be thrilled. To write my internet clients in Icon I have
>resorted to the UGLY expedient of writing a C program that opens a TCP
>connection, forces standard input and standard output to the
>connection, then exec's an Icon program. Of course, I can't easily
>write servers this way if they have peresistent state :-(
After observing this thread for a while.
Icon is, as we all know, good at data processing. Quite often if you rethink
your strategy you can arrange to have a small program generate the data and
then feed that data into Icon for processing. This is after all one of the
worthwile philosophies on which unix was based. Obviously this wouldn't be
suitable for all cases, especially where decisions had to be made on the
contents of the data for subsequent actions. Rather than try to reshape the
language for system tasks, perhaps we could simply extend the I/O interface so
that, given a suitable bit of code, we could open anything from a TCP socket
to a directory to whatever. If this all went through the open/read/write
interface there would be no loss of portability and no trying to change the
language away from its data processing style. There would probably need to
be an equivalent to ioctl as well.
Just a thought.
Eddie